Natural language processing applications using large language models

Abstract

The present disclosure relates to natural language processing applications using large language models. The methods presented herein can provide for the execution of specific types of tasks using large models without the need to retrain the model. Custom endpoints can be trained for specific types of tasks as can be indicated by the specification of one or more guidance mechanisms. Guidance mechanisms can be added to or used with requests to guide the model to perform a task type related to a text string. Endpoints receiving such requests can perform any orchestration required by the model to obtain the request in the format required by the model and can add the guidance mechanism to the request, for example, by adding one or more text strings (or text prefixes) in front of the text formatted request. The model receiving this string can process the text according to the guidance mechanism. Such methods can allow for a variety of tasks to be performed by a single model.

Description

Background Technology

[0001] Natural Language Processing (NLP) techniques are typically applied and implemented for a variety of tasks, such as language generation or analysis, grammar and usage checking, or content summarization, to name a few. To provide highly accurate NLP results, it can be advantageous to use large language models that have been trained on very large training sets of vocabulary and grammar. While large language models (LLMs) can be a very powerful and general-purpose tool for implementing a wide variety of complex NLP algorithms, the computational cost of such models is high, sometimes requiring many multiprocessor servers or workstations to load and perform the basic computations, making these models inaccessible to many potential users. While entities like cloud providers can host an LLM for a variety of different users or entities, such an implementation may be suboptimal. For example, for unique use cases or operations, the language model may need to be customized—requiring additional data specifically relevant to that use case or operation to train the model. Even for large cloud providers, the scale and computational cost of these models reduce the feasibility of training and hosting different large models for each such use case or operation. Attached Figure Description

[0002] Various embodiments according to this disclosure will be described with reference to the accompanying drawings, in which:

[0003] Figure 1 An example language reasoning processing system that can be used according to various embodiments is shown;

[0004] Figure 2 The diagram illustrates calls to different endpoints and different sizes of language models according to various embodiments;

[0005] Figure 3 Examples of suggested engineering and adjustments based on at least one embodiment are shown;

[0006] Figure 4 Different inference results based on different guidance mechanisms according to at least one embodiment are shown;

[0007] Figure 5 An example process for performing custom inference using a global model according to at least one embodiment is shown;

[0008] Figure 6 An example process for generating a custom endpoint for a specific type of task to be performed, according to at least one embodiment, is shown;

[0009] Figure 7 Components of an example distributed system, according to at least one embodiment, are shown;

[0010] Figure 8AThe inference and / or training logic according to at least one embodiment is illustrated;

[0011] Figure 8B The inference and / or training logic according to at least one embodiment is illustrated;

[0012] Figure 9 An example data center system according to at least one embodiment is shown;

[0013] Figure 10 A computer system according to at least one embodiment is shown;

[0014] Figure 11 A computer system according to at least one embodiment is shown;

[0015] Figure 12 At least a portion of a graphics processor according to one or more embodiments is shown;

[0016] Figure 13 At least a portion of a graphics processor according to one or more embodiments is shown;

[0017] Figure 14 This is an example data flow diagram of an advanced computing pipeline according to at least one embodiment;

[0018] Figure 15 This is a system diagram of an example system for training, adapting, instantiating, and deploying machine learning models in an advanced computing pipeline, according to at least one embodiment; and

[0019] Figure 16A and Figure 16B A data flow diagram of a process for training a machine learning model according to at least one embodiment is shown, as well as a client-server architecture for enhancing annotation tools using a pre-trained annotation model. Detailed Implementation

[0020] In the following description, various embodiments will be described. Specific configurations and details are presented for illustrative purposes to provide a thorough understanding of the embodiments. However, it will be apparent to those skilled in the art that the embodiments can be practiced without specific details. Furthermore, well-known features may be omitted or simplified so as not to obscure the embodiments being described.

[0021] Methods according to various embodiments can provide the use of large models for a variety of different tasks. In particular, various embodiments provide the ability to use a single large language model (LLM) for a variety of different natural language processing (NLP) related inference tasks without retraining the model or using multiple different, custom models. An inference service capable of using custom endpoints can be provided, and individual endpoints can be trained for specific types of tasks—this can be represented by selecting or providing one or more “guidance mechanisms.” Guidance mechanisms can include any token, label, weight, file, modifier, or other type of data object that can be added to, embedded in, or associated with a request to perform a specific operation or type of operation on a string of text. Example guidance mechanisms include a cue token indicating the type of operation to be performed, a retrieval set label indicating the dataset to be used for the operation, and / or adapter weights that can effectively modify the operation or structure of the language model, and other such options. An endpoint receiving such a request can perform orchestration and / or obtain other operations required by the request in the text format required by the language model, and guidance mechanisms can be added to the request, for example, by adding one or more text strings (or text prefixes) to the text format of the request. In some embodiments, there may be language model instances of different sizes, and individual endpoints may be associated with and trained on one of these specific-sized instances. This can affect aspects such as the time and cost of performing inference for a given request. Once inference is performed using the model, the results (in the form specified by a hint token) can be returned to the endpoint, where any orchestration and / or other processing can be performed to place the results in the necessary structure or format, and the results can be forwarded to the intended recipient or destination. Such a global service can be used by a variety of different users, developers, and / or other entities, each of whom can generate custom endpoints for a variety of different use cases or operation types. This heterogeneous set of tasks can be transformed into a highly homogeneous computational flow, or a sequence of homogeneous batch tasks, that can be processed using a single large language model, which can be highly efficient and computationally cost-effective.

[0022] The advantages of generating such representations can be found in a wide variety of applications and use cases. These can include, for example, but are not limited to, using them in conversational systems to understand the intent behind what a person is saying, enabling a digital conversational agent to respond appropriately. This will be broadly applicable to any situation where a computer system interacts with a person through spoken or written communication. This approach can also be used to analyze written documents or recordings of human language to determine their meaning, structure, or, for example, to find grammatical errors or generate alternative wording. This approach is also advantageous for generating written or spoken content, such as in video games, for marketing copy, or for creative writing or business communication applications. Another application is translation between different languages, such as translating from one language to another. Yet another application is in the context of computer programming source code, for systems that automatically generate code to perform a given task or analyze code to understand its structure, purpose, or detect programming errors. These language-related operations can correspond to a wide variety of applications or systems that may include one or more language-related tasks, such as those involving video conferencing, image analysis, autonomous navigation, robotics, games and animation, and data processing, among other such options.

[0023] Based on the teachings and suggestions contained herein, it will be apparent to those skilled in the art that variations of this and other such functions may also be used within the scope of various embodiments.

[0024] Figure 1An example system 100 is illustrated that can be used to perform various language-based tasks or operations. In this example, a user or entity can use a client device 150 (or other computerized device or system) to invoke (or request or instruct) system 100 to perform those tasks or operations. The invocation may include some amount of text to be processed, and specifications of one or more tokens (or other objects or files) indicating how the text should be processed. In some embodiments, the endpoint selected to receive the request may be associated with these one or more tokens, so that no separate instruction is required. In this example, a large language model will be used to process the text, which, as described herein, can be a powerful tool for performing text-based operations. Requests from client device 150 may be received in an environment such as a data center (or cloud computing environment or multi-tenant resource provider environment, etc.), where shared resources are used to host one or more large language models 104. In this example, these large language models 104 may be provided as part of a large language model (LLM) inference service 102 hosted by resources in the data center. Client device 150 can invoke appropriate endpoints 122 and 126, which can direct text to query service 106, which can then enable the text query or operation to be performed using appropriate LLM 104. The result of this processing, such as inference generated by the corresponding LLM, can then be returned to client device 150 or directed to another appropriate receiver or destination.

[0025] However, as mentioned earlier, while large language models (LLMs) are very powerful and general-purpose tools for implementing a wide range of complex NLP algorithms, previous LLM-based approaches have not scaled well, at least in part due to their corresponding computational costs. Providing a single LLM in previous systems would not yield accurate results due to the diverse use cases and operations that various users will offer, at least in part because it is impossible (or impractical) to train and host multiple independent LLMs for these different operations and use cases.

[0026] Therefore, various embodiments of this disclosure can propose a system architecture for one or more large language models hosted in a data center and accessed as a service by various users in a way that allows for flexible customization for various use cases and operations. Such an architecture also makes this deployment computationally efficient for the service provider. For example... Figure 1 The system 100 shown may include components such as LLM inference service 102; one or more inference namespaces 120, such as one namespace for each developer, user or entity; and training namespaces 140 for training endpoints for specific tasks, operations or use cases.

[0027] LLM inference service 102 can receive text-based requests and generate corresponding inference. In at least some embodiments, in the presence of LLM instances of different sizes (which may incur different usage costs), a request can specify the model to use, such as a smaller model (at least in one or more aspects of computation, memory, or storage requirements), which has lower hosting or operating costs and requires less time to perform computations, but is best suited for simple inference tasks; or a “large” model, which may have higher operating costs but provides more robust inference capabilities. While four different model sizes are shown, any number of LLM models 104 of various sizes may be used. There may also be other differences between model instances, such as those related to location or availability, and others of this kind. Requests or queries received by query service 106 from this inference service 102 can be directed to the appropriate model instance for inference, and the results (after any post-processing, aggregation, or other such operations) are returned to client device 150 or the target destination.

[0028] As described above, different requests can be associated with different tasks, operations, or use cases. Instead of training and hosting different models for these different choices, this example system can allow individual requests to be associated with one or more tags, tokens, weights, and / or other objects or files, which can help the inference service 102 determine how best to process the request or perform inference for the request. In this example, a request (at request or inference time) can be associated with one or more guidance mechanisms, such as at least one "hint token" (PT), "retrieval set tag" (RST), or "adapter weight" (AW), and other such options. In some embodiments, there can be a fixed set of guidance mechanism types, such as three different types, but it is not required that any given request provide three different types. These guidance mechanisms can allow for a variety of customized options without altering the computational flow of the inference computation through the LLM. As discussed in more detail herein, individual guidance mechanisms can cause the LLM to return different results, or result types, such as a complete sentence response, a true / false response, and / or the speech type of each word in a text string, and other such options, which may come from a specific set of knowledge or information. The way these mechanisms are handled allows LLM instances to use the same inference process or computation to process text input, so a single LLM can be used for all these customized requests. Furthermore, multiple requests with the same or different association mechanisms can be batched together for improved efficiency. Such an inference service 102 can host LLM instances of different sizes, such as those with 1 billion, 5 billion, 40 billion, and 530 billion parameters, ranging from small to large LLM instances 104.

[0029] To allow requests for different associated guidance mechanisms to undergo the same processing, various endpoints 122, 126 may be deployed, where individual endpoints may be associated with different guidance mechanisms or combinations of guidance mechanisms. In at least some embodiments where there may be multiple users, developers, or entities using the system, various managed inference namespaces 120 may be present, allowing each user (or developer, etc.) namespace for the deployed endpoints. A given inference namespace 120 may contain all endpoints for receiving requests to perform inference using the managed LLM 104 of the inference service 102. For example, these endpoints may be managed by the respective user or developer and, in some cases, may be exposed to other users or even the public (e.g., exposed to the open internet) for use. Each endpoint 122, 126 may consist of multiple associated items or aspects, such as a name, identifier, address, and / or Uniform Resource Locator (URL), such as " / summarize_news_article". The endpoint may also include an indicator of the size of the LLM, the request sent to that endpoint, and one or more associated inference parameters (e.g., sample count or temperature). Endpoints may also include, or be associated with, one or more guidance mechanisms 124, such as hint tokens and retrieval set tags. In at least some embodiments, endpoints may also be associated with the type of data orchestration 128 to be performed. In some embodiments, this may be an option provided to the user, developer, or entity that created the corresponding endpoint. Data orchestration choices may be made, specifying how input fields of a particular type (e.g., for structured language protocols such as JSON or Protobuf) are mapped to and out of text strings that will be processed by the LLM. Orchestration may be performed using code that executes within the endpoint or by a process that communicates with the endpoint. Requests for a user received from client device 150 (or other such sources) may be received using a load balancer or ingress layer 130, which may route the request to the appropriate endpoint 122, 126 in the namespace 120 associated with that user. For structured incoming requests, such as JSON requests received via a REST endpoint or Protobuf requests received via a GRPC endpoint, one or more orchestration components 128 or processes (possibly within the endpoint in some embodiments) can convert the structured request data into text according to the orchestration rules 142 used during the training and creation of the namespace 120 and / or the endpoint. The request can also be bound to any associated guidance mechanisms, such as hint tokens and retrieval set tags, and passed to the LLM inference service 102 for processing along with associated LLM parameters.Then, the reasoning-based response from the LLM inference service can be orchestrated back into a format based on the corresponding structured language protocol (such as JSON or Protobuf) according to one or more output data orchestration rules, and returned to the client device 150 or otherwise sent to the target recipient or destination.

[0030] For example, to provide users or developers with an appropriate set of endpoints configured for a specific set of tasks, such a system could also provide tools, such as a developer studio, that operate within the training namespace 140. Tools—such as a developer studio—could provide a workflow for training guidance mechanisms, such as cue tokens 144 or adapter weights, or for managing retrieval sets 146 that reference the LLM. One or more orchestration rules 142 may also be provided, as discussed herein.

[0031] Training can involve a specific training and evaluation dataset with training modules 148 for a given language model, and may require specifications for at least some training hyperparameters (e.g., learning rate or maximum number of training iterations). This training workflow can then perform multiple training iterations until final criteria are met, such as model convergence, loss values ​​below an acceptable threshold, and / or all training data has been used, among other such options. Users can perform tasks such as training and evaluation for each guidance mechanism intended for a given task.

[0032] Once customization and training are complete, the associated customized data can be published to a new endpoint in the corresponding inference namespace 120. The endpoint may have access control or other mechanisms or configurations in place, such as being provided as part of an endpoint management system. Such developer tools can run in various locations, such as in the cloud, as part of the inference system, and / or deployed on-premises at the user's location, among other such options.

[0033] If the tool is provided as part of a managed cloud service, if the underlying LLM is updated, the system can automatically recalculate the cue tokens or adapter weights from the associated training data and automatically redeploy the associated endpoints. At any time, users can download the associated guidance mechanisms and use them for inference at their location with the associated LLM without going through the inference namespace 120. The developer studio can also be configured to perform additional functions to save processing time. For example, endpoints can be trained for a given cue token, result set labels, and / or adapter weights relative to a small LLM. Since the training process doesn't require much additional processing when done together, the studio can also generate endpoints for the same guidance mechanisms, but for other sizes of LLMs, such as medium and large LLMs, or at least the next larger LLM. Creating new endpoints from scratch can be resource-intensive, so training and creating them together for model instances of different sizes can be advantageous. The developer studio can also be configured to automatically generate updated endpoints for any updates or modifications to the corresponding language model.

[0034] Figure 2 An exemplary call flow in such a system 200 is illustrated. In this example, there are two different client devices 202 and 204, which may be associated with the same entity or different entities, and they may call, as relative to... Figure 1The invocation can be made within the same or different namespaces discussed. In some embodiments, there may also be a generic or "public" namespace with a common hint token and / or a publicly available retrieval set that can be used by multiple different entities. In this example, client A 202 has a text string that the LLM wants to process using inference service 216. Based at least on the intended use of the processing result, the request may have one or more associated guidance mechanisms 218, such as a specific hint token for identifying the type of result to be returned, and a retrieval set for identifying the specific database that the LLM will use to process the request. In this example, there will be a trained endpoint, here endpoint A 204, which is specifically trained for the text associated with these specific guidance mechanisms 218. The endpoint can take any suitable form, such as a REST endpoint. There will typically be other endpoints, such as endpoint B 208, which may be in the same namespace, but will not be invoked for this request because they have been trained for different guidance mechanisms. The endpoints can also correspond to different types, so that one subset of the endpoints may be REST endpoints, while another subset may be endpoints of another type, whether in the same or different namespaces. The invoked endpoint, here endpoint A 206, can use these guidance tokens to transform the input request—which contains the text to be processed and the associated guidance mechanism—into a request formatted suitable for processing by the general LLM inference service 216. In some embodiments, the client may submit the request to a generic interface or API instead of invoking a specific endpoint; for example, it may analyze the request and direct it to the appropriate endpoint. Each endpoint may also be associated with, or trained to, a specific-sized LLM, such that endpoint A 206 can be trained, configured, or customized to send the formatted string to the LLM inference service 216 for processing by the large LLM 212. The inference result in this example can be returned to client A 202 via endpoint A 206, although in other examples, the result may be passed through different endpoints or sent to different (or additional) receivers or destinations. Although not shown, the processed result may also be stored by, or for, the LLM inference service for use in further training of the LLM model.

[0035] As shown in the figure, a single client device B 204 can submit a request including the text to be processed and a set of different guidance mechanisms 220, which will then be directed to different endpoints C 210 associated with these guidance mechanisms. If client B 204 and client A 202 are associated with a single entity, then these endpoints 206, 208, and 210 can all be associated with a single namespace; otherwise, at least endpoint A 206 and endpoint C 210 can be in different namespaces. In this example, endpoint C 210 then directs the formatted request to the target LLM, which can be any LLM, such as the small LLM 214 discussed, but in this example, it is also directed to the large LLM 212 because a single LLM model can be shared by multiple different entities for multiple different types of natural language processing (or other such) tasks. The result can be returned via endpoint C 210 (or another endpoint) and back to client device B 204 and / or another target destination or receiver. Although in Figure 2 Not shown in the document, but the endpoint (or the process communicating with the endpoint) can also perform actions regarding... Figure 1 The orchestration discussed here refers to a process where requests received from the client can be in a structured format, such as JSON, which can be converted to text to generate a text request to be provided to the LLM, and responses received from the LLM (and inference service) can be in a text format, or they can be orchestrated so that the responses received back to the requesting client device are in an appropriate structured format.

[0036] In at least one embodiment, a cue token may be used to provide guidance or instructions on how to process the associated text. The cue token may take the form of a series of numbers that can be placed before (or otherwise inserted into or after) the input request text string. This series of numbers (or alphanumeric characters) may not be human-readable, but it can provide guidance on how to process subsequent text within the string. The text string can be a sentence in natural language (e.g., American English). Regardless of the processing type used, the input to the LLM can be a natural language text string prefixed with an alphanumeric sequence indicating the type of reasoning to be performed or the type of response to be returned.

[0037] Figure 3Example phase 300 of the training and / or generation of cue tokens is shown. In this example, a training text input string 302 is provided, posing the question "In which country is Normandy located?" In cue engineering phase 304, a cue token corresponding to the instruction "Answer the question" is generated. Although the cue is displayed in human-readable text here, it should be understood that in many cases the cue token will be in the form of an alphanumeric string, which may not be human-readable. During training, the cue token generator can analyze training data associated with this task type, such as the input text and the corresponding cue token, to learn to generate appropriate cue tokens to complete the target task. As previously mentioned, there may be a cue token stored for each endpoint or associated with each endpoint, which in at least some embodiments may be the corresponding API, respectively. In addition to generating representative strings as cue tokens, cue engineering phase 304 may also identify and store routines for extracting results from the generated text.

[0038] The cues in this example can also undergo a cue refinement phase 310. The cue refinement phase can involve any process attempting to improve the results of LLM inference, where the LLM is “frozen” or fixed and generally cannot be customized for a given task. By improving the quality of the cue tokens and associated processes, more accurate results can be obtained for a specific task. In this example, the input task can be tokenized 314 to allow for more granular analysis of the components, which may help generate one or more “soft” cue tokens 312. Individual soft cue tokens 316 can modify things about the main cue token, or a set of soft cue tokens 312 can be used in place of a single cue token 306. Soft cue tokens 312 can take the form of virtual tokens, each token representing a more specific instruction. Instead of simply instructing the LLM to answer a question, soft tokens may indicate how to answer, the emotional state being attempted to convey, the form of the answer, and / or so on. These cues can take various forms, such as textual cues in some embodiments, embeddings learned from cue refinement, and / or learned cue models. The returned result can be in a specified form, such as a guide token, a single Boolean value, structured text, multiple responses—such as the speech type for each word in the input text string, and so on. In some embodiments, the result can also provide a determination made relative to the input text, such as the sentiment expressed by a movie review.

[0039] Since LLMs are text-only, any instructions provided must also be in text format. The cue token can take the form of a text prefix that provides some instructions or "guidance" on the reasoning to be performed. As mentioned earlier, this could be instructing the model to answer a question in a true / false manner, inferring the sentiment of text, determining whether a sentence is logically sound, extracting specific types of information (e.g., name, address, or title, if present), summarizing paragraphs, proposing topic lines for the body of text, or answering questions using a library specified by the retrieval set labels, among other such options. An LLM can be considered a model that understands natural language and can be asked to do many different things, as long as the request is also in that natural language and the model has been trained to understand the request. During training, the model can evaluate examples of the type of task to be performed and can generate sequences of numbers (or alphanumeric characters) based on a computer- or network-understandable language or syntax that the model can comprehend. The model's results can be evaluated, and if the results are inaccurate or not of the desired type, further training can be performed to try to refine the instruction string used for the cue token. This token can be refined using processes such as P-tuning.

[0040] Similar to other guidance mechanisms, adapter weights can be inserted into the text of a request to be sent to the inference service or language model, for example, by appending them to the beginning of the text string to be processed. In some embodiments, guidance mechanisms will be appended in a specific order, while in other embodiments, the order may not be fixed, but individual guidance mechanisms may include indicators (e.g., unique codes or symbols) that identify a portion of the text string corresponding to a particular guidance mechanism. Adapter weights can act as perturbations to the LLM activation, which can cause the LLM to return different types of results for different requests with different adapter weights. Adaptor weights can be thought of as adjustments to one or more network weights, which can effectively allow the network to infer in a slightly different way without having to retrain to adjust the relevant network weights. The training process, as discussed in [the text], [is as follows]. Figure 1 The discussed approach can be used to determine appropriate adapter weights for a given task, i.e., to associate them with the corresponding endpoints. Example adapter weights can indicate which additional processing layers should be placed between two layers of the model, such as between layers 37 and 38, with specific weights or activations. This method can effectively modify the model's operation without requiring retraining.

[0041] Retrieval set labels can be used to specify a database, dictionary, library, or other dataset or facts that will be used for a specific reasoning task associated with an endpoint. One advantage of using such labels is that it allows the language model to use data beyond what was used to train the model, effectively extending the model's capabilities without requiring retraining on that data. Two different requests can point to two different databases with different fact sets, so the model might answer the exact same question with two different answers. Everything else related to these requests, except for database lookups, can be computationally equivalent or homogeneous, which helps make this process very efficient. The format of the retrieval set can vary, as long as the data is in a text format that the relevant language model or reasoning service can understand and access. In some embodiments, a certain amount of preprocessing may be performed, where the data is extracted from the database or fact set indicated by the retrieval set label, and the data is formatted and indexed into a specific format to be provided as input to the language model. This approach allows for greater diversity in the specified database in structure or form, as long as the data itself can be placed in a single format for use by the language model. In some embodiments, preprocessing may efficiently index the fact set and provide that index to the model, as well as other such options, so that the language model or inner layers of the network can access those facts (or documents, etc.) identified through the index. Different datasets can be suitable for specific tasks. For example, a database of legal terms is more useful for different tasks than a set of recipes. Merging these facts into a single database generally offers no benefit; it would either become excessively large or require reducing the amount of facts of any given type to be practical, which would degrade the quality of the results. In some cases, there may be four distinct endpoints that are identical except that they use different sets of facts and are trained or customized for them.

[0042] Figure 4A set of different types of inference results 400 are shown, which can be returned for the same text string using the same language model but different custom endpoints. It should be understood that while “endpoint” is used here as the primary example, there can be any number of ways to apply specific or custom processing to an incoming request using various models, rules, identifiers, interfaces, and / or other such options, such that “endpoint” (whether RESTful or otherwise) should not be considered a limitation on the scope of various embodiments. In this example, a single text string 402 can be received by multiple different endpoints, which can be associated with different guidance mechanisms, or combinations of guidance mechanisms. As shown, different types of results 404 can be returned by using different prompt tokens, or other such guidance mechanisms. In this example, for the text string “Is the Earth round?”, the results could have similar content but different forms, such as “Yes” for a yes / no result, “True” for a true / false result, “The Earth is round” for a “Answer the question in natural language” result, or “1” for a binary 1 / 0 result. Another result of different prompt tokens returns the part-of-speech tag of the text string, at least determined by a specific set of definitions. Another result shows the difference in results based on a specified set of facts. For example, if asked whether the Earth is round, a modern set of facts might return a "yes" answer, but if someone determines the response based on a set of facts, such as medieval European religious texts, then result 406 might be a "no" yes / no result, because that is the correct answer based on that particular set of facts. As shown in the figure, by using different guidance mechanisms, a single large language model can be guided to return very different results.

[0043] This approach allows multiple users to efficiently customize the performance of a model, such as an LLM, without actually modifying or retraining the model. These users can then use custom endpoints that format text processing requests using one or more guidance mechanisms, so the LLM knows how to handle that specific request to perform a particular operation or task type. Multiple different users can then use such a model simultaneously to perform multiple different types of tasks. As mentioned earlier, for efficiency, these similarly formatted requests can also be batched together for transmission and processing, even if they may correspond to different types of tasks to be performed or results to be provided. The operations performed by the LLM may not change because of these guidance mechanisms, allowing a single model to perform the same (or similar) computations without retraining or customization, unless improvements are needed over time or for another such purpose. For example, if the retrieval set labels indicate the use of a specific database, the model will perform slightly different computations. However, it is possible to reduce tens of thousands of custom requests to a single computation to be performed, or a small set of computations (e.g., 2-10). Essentially, such a system or service, according to at least one embodiment, can receive a set of heterogeneous tasks that can be converted into homogeneous computational strings that can be processed by a single model or instances of the same model, which can make the system or service more efficient and less computationally expensive.

[0044] Figure 5An example process 500 for enabling a large language model to perform custom inference tasks that may be performed according to various embodiments is shown. It should be understood that, for this process and others discussed herein, within the scope of various embodiments, there may be additional, fewer, or alternative operations performed in a similar or alternative order, or at least partially in parallel, unless otherwise specifically stated. In this example, a request is received at 502 from an endpoint associated with a specific natural language processing (NLP) operation, such as a specific inference task to be performed to return a specific type of result. The request in this example may include structured text, such as JSON or similar formatted text. The structured text (if present) may be converted at 504 to unstructured text in a natural language format (or another text format used by the target language model), for example, by using an orchestration / unorthodoxy process specific to that endpoint. The endpoint may also be added as a prefix (or other additive) at 506 to the text string to be processed, with one or more alphanumeric prefixes corresponding to one or more guidance mechanisms. As discussed herein, this may include adding one or more alphanumeric strings that in some way instruct or modify the operation of the large language to obtain a specific type of inference result. The text request can be forwarded to a large language model (e.g., or an inference service hosting the model) for processing or inference, where the model to which the request is forwarded can be a specific-sized instance associated with an endpoint. The large language model can then perform inference on the text string in the request, as instructed by one or more guidance mechanisms. The inference result can then be received back to the endpoint, which can convert the text back to a structured format if necessary. The result can then be forwarded to a target recipient or destination, which may be different from or the same as the source of the request. As previously described, in some embodiments, the request can be batched with other requests or added to a single result stream, which helps improve operational efficiency.

[0045] Figure 6An example process 600 for generating task-specific endpoints for a trained model, which can be performed according to various embodiments, is illustrated. In this example, a request 602 (to a developer's studio or other such system or service) is received to generate a custom endpoint for a natural language task type to be performed using a large language model. For this endpoint, numerous operations can be performed, shown in sequence, but may also be performed simultaneously or in other suitable order. In one operation, it can be determined 604 whether there are any orchestration rules (or other formatting or processing rules) to be applied to format the incoming request into a target format for the language model, such as unstructured natural language. Other formats or rules may be available for other types of models or tasks. For a given type of task, a hint token 606 can be determined, which may indicate the type of inference to be performed and / or the type or result to be returned. In some cases, this may involve a generation and / or training process that provides a token that can guide the language model to perform the target task. The hint token can take any of many different forms, but for a language processing model, it may take the form of an alphanumeric string that provides code that the language model can understand in order to perform the target type of task. If a specific set of facts is to be used for a task, a 608 retrieval set label can be determined, which indicates to the model which set of facts should be used for inference. As mentioned earlier, such labels do not affect the processing to be performed, but rather indicate the dataset or content from which the model should extract during inference, such as determining the answer to a question. During this training process, it may also become clear whether some minor modifications to the operation of the language model are needed, or at least beneficial, where such modifications can be made by introducing one or more adapter weights, which can effectively modify one or more network weights or layers of the model. If so, these weights can be determined 610 or learned through the training process and provided for generating endpoints. In some cases, there will be a specific model, model type, and / or model size for its training endpoints; if so, this model can be determined 612 such that the endpoints are trained to work with that specific model instance. Using this information, 614 custom endpoints can be generated during this training or generation process, which can transform incoming requests according to provided orchestration (or other) rules and cause the language model's instruction instance to infer according to a determined guidance mechanism—for example, by adding a text string of the guidance mechanism to the request generated by the custom endpoint. 616 custom endpoints can then be provided to enable a large language model (or other such model or network) to perform specific types of inference or return specific types of results without having to retrain the language model or generate a new model. In at least some embodiments, the endpoint may be deployed to a namespace associated with the entity that generated the endpoint.

[0046] As discussed, the various methods presented in this paper are lightweight enough to be executed in real-time or near real-time on client devices such as personal computers or game consoles. Such processing can be performed on content generated on or received from an external source (e.g., streaming content received via at least one network) on that client device. This source can be any suitable source, such as a game console, streaming provider, third-party content provider, or other client devices and other similar options. In some cases, the processing and / or rendering of this content can be performed by one of these other devices, systems, or entities and then provided to the client device (or another such receiver) for presentation or other similar purposes.

[0047] As an example, Figure 7An example network configuration 700 is shown that can be used to provide, generate, modify, encode, and / or send content. In at least one embodiment, a client device 702 can generate or receive session content using components of a content application 704 on the client device 702 and data locally stored on the client device. In at least one embodiment, a content application 724 (e.g., a text processing application) executing on a content server 720 (e.g., a cloud server or edge server) can initiate a session associated with at least the client device 702, such as using a session manager and user data stored in a user database 734, and content 732 can be determined by a content manager 726. The content generation or management application 726 can generate or receive requests to perform NLP-related processing, which can be passed to a large language model 730 for processing, such as by providing an inference service hosted on a server. In some embodiments, the request can be directed to a custom endpoint 728 associated with a specific type of task, which can then be performed by the model. At least a portion of the generated content (e.g., the result of inference or data from a custom endpoint) can be transmitted to client device 702 using a suitable transmission manager 722 for delivery via download, streaming, or other such transmission channels. An encoder can be used to encode and / or compress at least some of the data before it is transmitted to client device 702. In at least one embodiment, client device 702 receiving such content can provide it to a corresponding content application 704, which may also include, or alternatively include, a content manager 710, a custom endpoint 712, or a developer studio 714 for generating custom endpoints, and other such options. A decoder can also be used to decode data received via network 740 for presentation via client device 702, such as presenting image or video content via display 706, and presenting audio such as sound and music via at least one audio playback device 708 such as a speaker or headphones. In at least one embodiment, at least some of the content may have already been stored on, rendered on, or made accessible to the client device 702, such that at least this portion of the content does not need to be transmitted over the network 740, for example, in cases where the content may have been previously downloaded or locally stored on a hard drive or optical disc. In at least one embodiment, a transmission mechanism such as data streaming may be used to transmit the content from the server 720 or user database 734 to the client device 702. In at least one embodiment, at least a portion of the content may be obtained from or streamed from another source, such as a third-party service 760, which may also include a content application 762 for generating or providing the content.In at least one embodiment, portions of the function may be performed using multiple computing devices or multiple processors within one or more computing devices, such as a combination of CPU and GPU.

[0048] In this example, these client devices can include any suitable computing device, such as desktop computers, laptops, set-top boxes, streaming devices, game consoles, smartphones, tablets, VR headsets, AR goggles, MR headsets / goggles, wearable computers, or smart TVs. Each client device can submit requests via at least one wired or wireless network, which can include the Internet, Ethernet, a local area network (LAN), or a cellular network, among other such options. In this example, these requests can be submitted to an address associated with a cloud provider that operates or controls one or more electronic resources within a cloud provider environment, such as a data center or server cluster. In at least one embodiment, the request can be received or processed by at least one edge server located at the network edge and outside at least one security layer associated with the cloud provider environment. In this way, latency can be reduced by enabling client devices to interact with servers in closer proximity, while also improving the security of resources within the cloud provider environment.

[0049] In at least one embodiment, such a system can be used to perform graphics rendering operations. In other embodiments, such a system can be used for other purposes, such as providing image or video content to test or validate autonomous machine applications, or for performing deep learning operations. In at least one embodiment, such a system can be implemented using edge devices, or can be combined with one or more virtual machines (VMs). In at least one embodiment, such a system can be implemented at least partially in a data center or at least partially using cloud computing resources.

[0050] Reasoning and training logic

[0051] Figure 8A Inference and / or training logic 815 is shown for performing inference and / or training operations associated with one or more embodiments. The following is in conjunction with... Figure 8A and / or Figure 8B Provide details about reasoning and / or training logic 815.

[0052] In at least one embodiment, inference and / or training logic 815 may include, but is not limited to, code and / or data storage 801 for storing forward and / or output weights and / or input / output data, and / or other parameters configuring neurons or layers of a neural network trained for and / or used for inference in one or more embodiments. In at least one embodiment, training logic 815 may include or be coupled to code and / or data storage 801 for storing graph code or other software to control timing and / or sequence, wherein weight and / or other parameter information is loaded to configure logic, including integer and / or floating-point units (collectively referred to as Arithmetic Logic Units (ALUs)). In at least one embodiment, code (such as graph code) loads weight or other parameter information into the processor ALU based on the architecture of the neural network to which the code corresponds. In at least one embodiment, code and / or data storage 801 stores weight parameters and / or input / output data of each layer of a neural network trained or used in one or more embodiments during forward propagation of input / output data and / or weight parameters during training and / or inference using one or more embodiments. In at least one embodiment, any portion of the code and / or data storage 801 may be included within other on-chip or off-chip data storage, including the processor's L1, L2, or L3 cache or system memory.

[0053] In at least one embodiment, any portion of the code and / or data storage 801 may be internal or external to one or more processors or other hardware logic devices or circuits. In at least one embodiment, the code and / or data storage 801 may be a cache memory, dynamic random access memory (“DRAM”), static random access memory (“SRAM”), non-volatile memory (e.g., flash memory), or other storage. In at least one embodiment, the choice of whether the code and / or data storage 801 is internal or external to the processor, for example, or composed of DRAM, SRAM, flash memory, or some other storage type, may depend on the available on-chip or off-chip storage space, the latency requirements of the training and / or inference functions being performed, the batch size of the data used in the inference and / or training of the neural network, or some combination of these factors.

[0054] In at least one embodiment, inference and / or training logic 815 may include, but is not limited to, code and / or data storage 805 for storing backpropagation and / or output weights and / or input / output data corresponding to neurons or layers of a neural network trained and / or used for inference in one or more embodiments. In at least one embodiment, during training and / or inference using one or more embodiments, code and / or data storage 805 stores weight parameters and / or input / output data for each layer of a neural network trained or used in one or more embodiments during backpropagation of input / output data and / or weight parameters. In at least one embodiment, training logic 815 may include or be coupled to code and / or data storage 805 for storing graph code or other software to control timing and / or sequence, wherein weight and / or other parameter information is loaded to configure logic including integer and / or floating-point units (collectively, an arithmetic logic unit (ALU)). In at least one embodiment, code (such as graph code) loads weight or other parameter information into the processor ALU based on the architecture of the neural network to which the code corresponds. In at least one embodiment, any portion of the code and / or data storage 805 may be included together with other on-chip or off-chip data storage, including the processor's L1, L2, or L3 cache or system memory. In at least one embodiment, any portion of the code and / or data storage 805 may be internal or external to one or more processors or other hardware logic devices or circuits. In at least one embodiment, the code and / or data storage 805 may be cache memory, DRAM, SRAM, non-volatile memory (e.g., flash memory), or other storage. In at least one embodiment, the choice between the code and / or data storage 805 being internal or external to the processor, for example, whether it consists of DRAM, SRAM, flash memory, or some other type of storage, depends on whether the available storage is on-chip or off-chip, the latency requirements of the training and / or inference functions being performed, the data batch size used in the inference and / or training of the neural network, or some combination of these factors.

[0055] In at least one embodiment, code and / or data storage 801 and code and / or data storage 805 may be separate storage structures. In at least one embodiment, code and / or data storage 801 and code and / or data storage 805 may be the same storage structure. In at least one embodiment, code and / or data storage 801 and code and / or data storage 805 may be partially identical and partially separate storage structures. In at least one embodiment, any portion of code and / or data storage 801 and code and / or data storage 805 may be included with other on-chip or off-chip data storage, including the processor's L1, L2, or L3 cache or system memory.

[0056] In at least one embodiment, the inference and / or training logic 815 may include, but is not limited to, one or more arithmetic logic units (“ALUs”) 810 (including integer and / or floating-point units) for performing logical and / or mathematical operations at least in part based on or instructed by training and / or inference code (e.g., graph code), the results of which may produce activations (e.g., output values ​​from layers or neurons within a neural network) stored in activation storage 820, which are functions of input / output and / or weight parameter data stored in code and / or data storage 801 and / or code and / or data storage 805. In at least one embodiment, activation is activated in response to execution instructions or other code, and linear algebraic and / or matrix-based mathematical generation performed by ALU 810 is stored in activation storage 820, wherein weight values ​​stored in code and / or data storage 805 and / or code and / or data storage 801 are used as operands with other values, such as bias values, gradient information, momentum values, or other parameters or hyperparameters, and any or all of these can be stored in code and / or data storage 805 or code and / or data storage 801 or other on-chip or off-chip storage.

[0057] In at least one embodiment, one or more processors or other hardware logic devices or circuits include one or more ALUs 810, while in another embodiment, one or more ALUs 810 may be located outside the processor or other hardware logic device or the circuitry that uses them (e.g., a coprocessor). In at least one embodiment, one or more ALUs 810 may be included within an execution unit of a processor, or otherwise included in a group of ALUs accessible by the execution unit of the processor, which may be within the same processor or distributed among different processors of different types (e.g., a central processing unit, a graphics processing unit, a fixed-function unit, etc.). In at least one embodiment, code and / or data storage 801, code and / or data storage 805, and activation storage 820 may be on the same processor or other hardware logic device or circuitry, while in another embodiment, they may be on different processors or other hardware logic devices or circuitries, or in some combination of the same and different processors or other hardware logic devices or circuitries. In at least one embodiment, any portion of activation storage 820 may be included together with other on-chip or off-chip data storage, including the processor's L1, L2, or L3 cache or system memory. Furthermore, inference and / or training code may be stored together with other code accessible to the processor or other hardware logic or circuitry, and may be retrieved and / or processed using the processor’s fetch, decode, schedule, execute, exit, and / or other logic circuitry.

[0058] In at least one embodiment, the active memory 820 may be a cache memory, DRAM, SRAM, non-volatile memory (e.g., flash memory), or other memory. In at least one embodiment, the active memory 820 may be wholly or partially located inside or outside one or more processors or other logic circuits. In at least one embodiment, the choice of whether the active memory 820 is internal to or external to the processor may depend on the available on-chip or off-chip storage, the latency requirements for training and / or inference functions, the batch size of data used in inference and / or training the neural network, or some combination of these factors. For example, it may include DRAM, SRAM, flash memory, or other memory types. In at least one embodiment, Figure 8A The inference and / or training logic 815 shown can be used in conjunction with an application-specific integrated circuit (“ASIC”), such as those from Google. Processing unit, from Graphcore TM Inference processing unit (IPU) or from Intel Corp. (e.g., "LakeCrest") processor. In at least one embodiment, Figure 8A The inference and / or training logic 815 shown can be used in conjunction with central processing unit (“CPU”) hardware, graphics processing unit (“GPU”) hardware, or other hardware such as field programmable gate array (“FPGA”)

[0059] Figure 8B Inference and / or training logic 815 according to at least one or more embodiments is illustrated. In at least one embodiment, the inference and / or training logic 815 may include, but is not limited to, hardware logic, wherein computational resources are dedicated or otherwise uniquely used in conjunction with weight values ​​or other information corresponding to one or more layers of neurons within a neural network. In at least one embodiment, Figure 8B The inference and / or training logic 815 shown can be used in conjunction with an application-specific integrated circuit (ASIC), such as those from Google. Processing unit, from Graphcore TM Inference processing unit (IPU) or from Intel Corp. (e.g., "LakeCrest") processor. In at least one embodiment, Figure 8BThe inference and / or training logic 815 shown can be used in conjunction with central processing unit (CPU) hardware, graphics processing unit (GPU) hardware, or other hardware (e.g., field-programmable gate array (FPGA)). In at least one embodiment, the inference and / or training logic 815 includes, but is not limited to, code and / or data storage 801 and code and / or data storage 805, which can be used to store code (e.g., graph code), weight values, and / or other information, including bias values, gradient information, momentum values, and / or other parameter or hyperparameter information. Figure 8B In at least one embodiment shown, each of code and / or data storage 801 and code and / or data storage 805 is associated with dedicated computing resources (e.g., computing hardware 802 and computing hardware 806), respectively. In at least one embodiment, each of computing hardware 802 and computing hardware 806 includes one or more ALUs that perform mathematical functions (e.g., linear algebraic functions) only on the information stored in code and / or data storage 801 and code and / or data storage 805, respectively, and the results of the function execution are stored in activation memory 820.

[0060] In at least one embodiment, each of the code and / or data storage 801 and 805 and the corresponding computing hardware 802 and 806 corresponds to a different layer of the neural network, such that activation obtained from one “store / computation pair 801 / 802” of the code and / or data storage 801 and computing hardware 802 provides input as input to the next “store / computation pair 805 / 806” of the code and / or data storage 805 and computing hardware 806, in order to reflect the conceptual organization of the neural network. In at least one embodiment, each store / computation pair 801 / 802 and 805 / 806 may correspond to more than one neural network layer. In at least one embodiment, additional store / computation pairs (not shown) may be included in the inference and / or training logic 815 after or in parallel with the store / computation pairs 801 / 802 and 805 / 806.

[0061] Data Center

[0062] Figure 9 An example data center 900 that can be used with at least one embodiment is shown. In at least one embodiment, the data center 900 includes a data center infrastructure layer 910, a framework layer 920, a software layer 930, and an application layer 940.

[0063] In at least one embodiment, such as Figure 9As shown, the data center infrastructure layer 910 may include a resource coordinator 912, grouped computing resources 914, and node computing resources (“nodes CR”) 916(1)-916(N), where “N” represents any positive integer. In at least one embodiment, nodes CR 916(1)-916(N) may include, but are not limited to, any number of central processing units (“CPUs”) or other processors (including accelerators, field-programmable gate arrays (FPGAs), graphics processors, etc.), memory devices (e.g., dynamic read-only memory), storage devices (e.g., solid-state drives or disk drives), network input / output (“NWI / O”) devices, network switches, virtual machines (“VMs”), power modules, and cooling modules, etc. In at least one embodiment, one or more nodes CR 916(1)-916(N) may be servers having one or more of the aforementioned computing resources.

[0064] In at least one embodiment, the grouped computing resources 914 may include individual groups (not shown) of node CRs housed in one or more racks, or a plurality of racks (also not shown) housed in data centers in various geographic locations. The individual groups of node CRs within the grouped computing resources 914 may include computing, networking, memory, or storage resources that can be configured or allocated to support groups of one or more workloads. In at least one embodiment, several node CRs, including CPUs or processors, may be grouped within one or more racks to provide computing resources to support one or more workloads. In at least one embodiment, the one or more racks may also include any number of power modules, cooling modules, and network switches, in any combination.

[0065] In at least one embodiment, resource coordinator 912 may configure or otherwise control one or more nodes CR916(1)-916(N) and / or grouped computing resources 914. In at least one embodiment, resource coordinator 912 may include a software design infrastructure (“SDI”) management entity for data center 900. In at least one embodiment, resource coordinator may include hardware, software, or some combination thereof.

[0066] In at least one embodiment, such as Figure 9As shown, framework layer 920 includes job scheduler 922, configuration manager 924, resource manager 926, and distributed file system 928. In at least one embodiment, framework layer 920 may include a framework of software 932 supporting software layer 930 and / or one or more applications 942 of application layer 940. In at least one embodiment, software 932 or application 942 may respectively include web-based service software or applications, such as services or applications provided by Amazon Web Services, Google Cloud, and Microsoft Azure. In at least one embodiment, framework layer 920 may be, but is not limited to, a free and open-source software web application framework, such as Apache Spark™ (hereinafter referred to as "Spark") which can be used for large-scale data processing (e.g., "big data") with distributed file system 928. In at least one embodiment, job scheduler 922 may include Spark drivers to facilitate the scheduling of workloads supported by the various layers of data center 900. In at least one embodiment, configuration manager 924 may be able to configure different layers, such as software layer 930 and framework layer 920 including Spark and distributed file system 928 for supporting large-scale data processing. In at least one embodiment, resource manager 926 is capable of managing cluster or group computing resources mapped to or allocated to support distributed file system 928 and job scheduler 922. In at least one embodiment, cluster or group computing resources may include group computing resources 914 on data center infrastructure layer 910. In at least one embodiment, resource manager 926 may coordinate with resource coordinator 912 to manage these mapped or allocated computing resources.

[0067] In at least one embodiment, the software 932 included in the software layer 930 may include software used by at least a portion of the nodes CR916(1)-916(N), the grouped computing resources 914, and / or the distributed file system 928 of the framework layer 920. One or more types of software may include, but are not limited to, Internet web page search software, email virus scanning software, database software, and streaming video content software.

[0068] In at least one embodiment, the application layer 940 may include one or more applications 942 that can be used by at least a portion of nodes CR916(1)-916(N), grouped computing resources 914, and / or the distributed file system 928 of the framework layer 920. The one or more types of applications may include, but are not limited to, any number of genomics applications, cognitive computing and machine learning applications, including training or inference software, machine learning framework software (e.g., PyTorch, TensorFlow, Caffe, etc.), or other machine learning applications used in conjunction with one or more embodiments.

[0069] In at least one embodiment, any of the configuration manager 924, resource manager 926, and resource coordinator 912 can implement any number and type of self-modification actions based on any amount and type of data acquired in any technically feasible manner. In at least one embodiment, self-modification actions can mitigate potentially poor configuration decisions by data center operators of data center 900 and can prevent underutilization and / or poor performance of the data center.

[0070] In at least one embodiment, data center 900 may include tools, services, software, or other resources to train one or more machine learning models or to use one or more machine learning models to predict or infer information according to one or more embodiments described herein. For example, in at least one embodiment, a machine learning model can be trained by calculating weight parameters based on a neural network architecture using the software and computing resources described above with respect to data center 900. In at least one embodiment, information can be inferred or predicted using trained machine learning models corresponding to one or more neural networks using the resources described above with respect to data center 900 by using weight parameters calculated through one or more training techniques described herein.

[0071] In at least one embodiment, the data center may use a CPU, application-specific integrated circuit (ASIC), GPU, FPGA, or other hardware to utilize the aforementioned resources to perform training and / or inference. Furthermore, one or more of the aforementioned software and / or hardware resources may be configured as a service to allow a user to train or perform information inference, such as image recognition, speech recognition, or other artificial intelligence services.

[0072] Inference and / or training logic 815 is used to perform inference and / or training operations associated with one or more embodiments. This document combines... Figure 8A and / or Figure 8BDetails are provided regarding the inference and / or training logic 815. In at least one embodiment, the inference and / or training logic 815 can be implemented in the system. Figure 9 Used in systems for reasoning or predicting operations based at least in part on weight parameters calculated using neural network training operations, neural network functions and / or architectures or neural network use cases described herein.

[0073] Such components can be used to support the execution of custom tasks using the same model (such as a large language model) without the need to retrain the model.

[0074] Computer System

[0075] Figure 10 This is a block diagram illustrating an exemplary computer system according to at least one embodiment. The exemplary computer system may be a system of interconnected devices and components, a system-on-a-chip (SoC), or some combination thereof 1000 formed with a processor, which may include an execution unit to execute instructions. In at least one embodiment, according to this disclosure, such as the embodiments described herein, the computer system 1000 may include, but is not limited to, components such as processor 1002, whose execution unit includes logic to execute algorithms for process data. In at least one embodiment, the computer system 1000 may include a processor, such as those available from Intel Corporation of Santa Clara, California. Processor family, Xeon™ XScale™ and / or StrongARM™ Core TM or Nervana TM A microprocessor may be used, although other systems (including PCs, engineering workstations, set-top boxes, etc.) with other microprocessors may also be used. In at least one embodiment, the computer system 1000 may execute a version of the Windows operating system available from Microsoft Corporation of Redmond, Washington, although other operating systems (such as UNIX and Linux), embedded software, and / or graphical user interfaces may also be used.

[0076] The embodiments can be used in other devices, such as handheld devices and embedded applications. Some examples of handheld devices include cellular phones, Internet Protocol (IP) devices, digital cameras, personal digital assistants (“PDAs”), and handheld PCs. In at least one embodiment, the embedded application may include a microcontroller, a digital signal processor (“DSP”), a system-on-a-chip (SoC), a network computer (“NetPC”), a set-top box, a network hub, a wide area network (“WAN”) switch, or any other system that can execute one or more instructions according to at least one embodiment.

[0077] In at least one embodiment, the computer system 1000 may include, but is not limited to, a processor 1002, which may include, but is not limited to, one or more execution units 1008, to perform machine learning model training and / or inference according to the techniques described herein. In at least one embodiment, the computer system 1000 is a single-processor desktop or server system, but in another embodiment, the computer system 1000 may be a multiprocessor system. In at least one embodiment, the processor 1002 may include, but is not limited to, a Complex Instruction Set Computer (“CISC”) microprocessor, a Reduced Instruction Set Computing (“RISC”) microprocessor, a Very Long Instruction Word (“VLIW”) microprocessor, a processor implementing instruction set combination, or any other processor device, such as a digital signal processor. In at least one embodiment, the processor 1002 may be coupled to a processor bus 1010, which can transmit data signals between the processor 1002 and other components in the computer system 1000.

[0078] In at least one embodiment, processor 1002 may include, but is not limited to, a Level 1 (“L1”) internal cache memory (“cache”) 1004. In at least one embodiment, processor 1002 may have a single internal cache or multiple levels of internal cache. In at least one embodiment, the cache memory may reside external to processor 1002. Depending on specific implementation and requirements, other embodiments may also include a combination of internal and external caches. In at least one embodiment, register file 1006 may store different types of data in various registers, including but not limited to integer registers, floating-point registers, status registers, and instruction pointer registers.

[0079] In at least one embodiment, a logic execution unit 1008, which performs integer and floating-point operations, is also located within the processor 1002. In at least one embodiment, the processor 1002 may also include a microcode (“ucode”) read-only memory (“ROM”) for storing microcode of certain macro instructions. In at least one embodiment, the execution unit 1008 may include logic for processing a packaged instruction set 1009. In at least one embodiment, by including the packaged instruction set 1009 in the instruction set of the general-purpose processor 1002, along with the associated circuitry for executing the instructions, the packaged data in the processor 1002 can be used to perform operations used by numerous multimedia applications. In one or more embodiments, the execution of numerous multimedia applications can be accelerated and more efficiently performed by using the full width of the processor's data bus to perform operations on the packaged data, which may eliminate the need to transfer smaller data units on the processor's data bus to perform one or more operations on one data element at a time.

[0080] In at least one embodiment, the execution unit 1008 may also be used in a microcontroller, embedded processor, graphics device, DSP, and other types of logic circuitry. In at least one embodiment, the computer system 1000 may include, but is not limited to, memory 1020. In at least one embodiment, memory 1020 may be implemented as a dynamic random access memory (“DRAM”) device, a static random access memory (“SRAM”) device, a flash memory device, or other storage device. In at least one embodiment, memory 1020 may store instructions 1019 and / or data 1021 represented by data signals that can be executed by processor 1002.

[0081] In at least one embodiment, the system logic chip may be coupled to the processor bus 1010 and the memory 1020. In at least one embodiment, the system logic chip may include, but is not limited to, a memory controller hub (“MCH”) 1016, and the processor 1002 may communicate with the MCH 1016 via the processor bus 1010. In at least one embodiment, the MCH 1016 may provide a high-bandwidth memory path 1018 to the memory 1020 for instruction and data storage, as well as for storage of graphics commands, data, and textures. In at least one embodiment, the MCH 1016 may initiate data signals between the processor 1002, the memory 1020, and other components in the computer system 1000, and bridge data signals between the processor bus 1010, the memory 1020, and the system I / O 1022. In at least one embodiment, the system logic chip may provide a graphics port for coupling to a graphics controller. In at least one embodiment, MCH 1016 can be coupled to memory 1020 via high-bandwidth memory path 1018, and graphics / video card 1012 can be coupled to MCH 1016 via Accelerated Graphics Port (“AGP”) interconnect 1014.

[0082] In at least one embodiment, the computer system 1000 may use system I / O 1022, which is a proprietary hub interface bus, to couple MCH 1016 to I / O controller hub (“ICH”) 1030. In at least one embodiment, ICH 1030 may provide direct connectivity to certain I / O devices via a local I / O bus. In at least one embodiment, the local I / O bus may include, but is not limited to, a high-speed I / O bus for connecting peripheral devices to memory 1020, chipset, and processor 1002. Examples may include, but are not limited to, an audio controller 1029, a firmware hub (“FlashBIOS”) 1028, a wireless transceiver 1026, a data storage 1024, a conventional I / O controller 1023 including a user input and keyboard interface 1025, a serial expansion port 1027 (e.g., a Universal Serial Bus (USB) port), and a network controller 1034. Data storage 1024 may include a hard disk drive, floppy disk drive, CD-ROM device, flash memory device, or other mass storage device.

[0083] In at least one embodiment, Figure 10 A system including interconnected hardware devices or "chips" is shown, while in other embodiments, Figure 10An exemplary system-on-a-chip (SoC) may be illustrated. In at least one embodiment, the device may be interconnected with a proprietary interconnect, a standardized interconnect (e.g., PCIe), or some combination thereof. In at least one embodiment, one or more components of the computer system 1000 are interconnected using a compute fast link (CXL) interconnect.

[0084] The inference and / or training logic 815 is used to perform inference and / or training operations related to one or more embodiments. (The following is in conjunction with...) Figure 8A and / or Figure 8B Details are provided regarding the inference and / or training logic 815. In at least one embodiment, the inference and / or training logic 815 can be... Figure 10 Used in systems for reasoning or predicting operations based at least in part on weight parameters calculated using neural network training operations, neural network functions and / or architectures or neural network use cases described herein.

[0085] Such components can be used to support the execution of custom tasks using the same model (such as a large language model) without the need to retrain the model.

[0086] Figure 11 This is a block diagram illustrating an electronic device 1100 for utilizing processor 1110 according to at least one embodiment. In at least one embodiment, electronic device 1100 may be, for example, but not limited to, a laptop computer, tower server, rack server, blade server, desktop computer, tablet computer, mobile device, telephone, embedded computer, or any other suitable electronic device.

[0087] In at least one embodiment, system 1100 may include, but is not limited to, processor 1110 communicatively coupled to any suitable number or type of components, peripherals, modules, or devices. In at least one embodiment, processor 1110 uses a bus or interface coupling, such as an I2C bus, system management bus (“SMBus”), low pin count (LPC) bus, serial peripheral interface (“SPI”), high-definition audio (“HDA”) bus, serial advanced technology accessory (“SATA”) bus, universal serial bus (“USB”) (versions 1, 2, and 3), or universal asynchronous receiver / transmitter (“UART”) bus. In at least one embodiment, Figure 11 The system shown includes interconnected hardware devices or "chips," while in other embodiments, Figure 11 An exemplary system-on-a-chip (SoC) can be illustrated. In at least one embodiment, Figure 11 The device shown can be interconnected with proprietary interconnects, standardized interconnects (e.g., PCIe), or some combination thereof. In at least one embodiment, Figure 11One or more components are interconnected using Computational Fast Link (CXL) interconnects.

[0088] In at least one embodiment, Figure 11 It may include a display 1124, a touch screen 1125, a touchpad 1130, a near field communication unit (“NFC”) 1145, a sensor hub 1140, a thermal sensor 1146, a fast chipset (“EC”) 1135, a trusted platform module (“TPM”) 1138, a BIOS / firmware / flash (“BIOS, FWFlash”) 1122, a DSP 1160, a drive 1120 (e.g., a solid-state drive (“SSD”) or a hard disk drive (“HDD”)), a wireless local area network unit (“WLAN”) 1150, a Bluetooth unit 1152, a wireless wide area network unit (“WWAN”) 1156, a global positioning system (GPS) 1155, a camera (“USB 3.0 camera”) 1154 (e.g., a USB 3.0 camera) and / or a low-power double data rate (“LPDDR”) memory unit (“LPDDR3”) 1115 implemented in, for example, the LPDDR3 standard. These components can each be implemented in any suitable way.

[0089] In at least one embodiment, other components may be communicatively coupled to processor 1110 via the components described above. In at least one embodiment, accelerometer 1141, ambient light sensor (“ALS”) 1142, compass 1143, and gyroscope 1144 may be communicatively coupled to sensor hub 1140. In at least one embodiment, thermal sensor 1139, fan 1137, keyboard 1136, and touchpad 1130 may be communicatively coupled to EC 1135. In at least one embodiment, speaker 1163, earphone 1164, and microphone (“mic”) 1165 may be communicatively coupled to audio unit (“audio codec and Class D amplifier”) 1162, which in turn may be communicatively coupled to DSP 1160. In at least one embodiment, audio unit 1162 may include, for example, but not limited to, audio encoder / decoder (“codec”) and Class D amplifier. In at least one embodiment, SIM card (“SIM”) 1157 may be communicatively coupled to WWAN unit 1156. In at least one embodiment, components such as WLAN unit 1150, Bluetooth unit 1152, and WWAN unit 1156 can be implemented as next-generation form factor (NGFF).

[0090] The inference and / or training logic 815 is used to perform inference and / or training operations associated with one or more embodiments. (The following is in conjunction with...) Figure 8A and / or Figure 8BDetails are provided regarding the inference and / or training logic 815. In at least one embodiment, the inference and / or training logic 815 can be... Figure 10 The system is used to infer or predict operations based at least in part on weight parameters calculated using neural network training operations, neural network functions and / or architectures or neural network use cases described herein.

[0091] Such components can be used to support the execution of custom tasks using the same model (such as a large language model) without the need to retrain the model.

[0092] Figure 12 This is a block diagram of a processing system according to at least one embodiment. In at least one embodiment, system 1200 includes one or more processors 1202 and one or more graphics processors 1208, and may be a single-processor desktop system, a multi-processor workstation system, or a server system having a large number of processors 1202 or processor cores 1207. In at least one embodiment, system 1200 is a processing platform incorporated within a system-on-a-chip (SoC) integrated circuit for use in mobile, handheld, or embedded devices.

[0093] In at least one embodiment, system 1200 may include or be integrated into a server-based gaming platform, including a game console, mobile game console, handheld game console, or online game console, which are game and media consoles. In at least one embodiment, system 1200 is a mobile phone, smartphone, tablet computing device, or mobile internet device. In at least one embodiment, processing system 1200 may also include components coupled to or integrated into a wearable device, such as a smartwatch, smart glasses, augmented reality, or virtual reality device. In at least one embodiment, processing system 1200 is a television or set-top box device having one or more processors 1202 and a graphical interface generated by one or more graphics processors 1208.

[0094] In at least one embodiment, each of the one or more processors 1202 includes one or more processor cores 1207 for processing instructions that, when executed, perform operations against the system and user software. In at least one embodiment, each of the one or more processor cores 1207 is configured to process a particular instruction set 1209. In at least one embodiment, the instruction set 1209 may facilitate Complex Instruction Set Computing (CISC), Reduced Instruction Set Computing (RISC), or computation via Very Long Instruction Word (VLIW). In at least one embodiment, each processor core 1207 may process a different instruction set 1209, which may include instructions that facilitate emulation of other instruction sets. In at least one embodiment, the processor core 1207 may also include other processing devices, such as a digital signal processor (DSP).

[0095] In at least one embodiment, processor 1202 includes cache memory 1204. In at least one embodiment, processor 1202 may have a single internal cache or multiple levels of internal caches. In at least one embodiment, the cache memory is shared among various components of processor 1202. In at least one embodiment, processor 1202 also uses an external cache (e.g., a Level 3 (L3) cache or a last-level cache (LLC)) (not shown), which can be shared among processor cores 1207 using known cache coherence techniques. In at least one embodiment, processor 1202 further includes a register file 1206, which may include different types of registers for storing different types of data (e.g., integer registers, floating-point registers, status registers, and instruction pointer registers). In at least one embodiment, register file 1206 may include general-purpose registers or other registers.

[0096] In at least one embodiment, one or more processors 1202 are coupled to one or more interface buses 1210 to transmit communication signals, such as address, data, or control signals, between the processor 1202 and other components in the system 1200. In at least one embodiment, the interface bus 1210 may be a processor bus, such as a version of the Direct Media Interface (DMI) bus. In at least one embodiment, the interface bus 1210 is not limited to the DMI bus and may include one or more peripheral component interconnect buses (e.g., PCI, PCI Express), memory buses, or other types of interface buses. In at least one embodiment, the processor 1202 includes an integrated memory controller 1216 and a platform controller hub 1230. In at least one embodiment, the memory controller 1216 facilitates communication between memory devices and other components of the processing system 1200, while the platform controller hub (PCH) 1230 provides connectivity to I / O devices via a local I / O bus.

[0097] In at least one embodiment, memory device 1220 may be a dynamic random access memory (DRAM) device, a static random access memory (SRAM) device, a flash memory device, a phase-change memory device, or a device with suitable performance for use as processor memory. In at least one embodiment, memory device 1220 may be used as system memory of processing system 1200 to store data 1222 and instructions 1221 for use when one or more processors 1202 execute an application or process. In at least one embodiment, memory controller 1216 is also coupled to an optional external graphics processor 1212, which may communicate with one or more graphics processors 1208 of processor 1202 to perform graphics and media operations. In at least one embodiment, display device 1211 may be connected to processor 1202. In at least one embodiment, display device 1211 may include one or more internal display devices, such as in mobile electronic devices or laptop devices, or external display devices connected via a display interface (e.g., DisplayPort). In at least one embodiment, the display device 1211 may include a head-mounted display (HMD), such as a stereoscopic display device for virtual reality (VR) or augmented reality (AR) applications.

[0098] In at least one embodiment, the platform controller hub 1230 enables peripheral devices to connect to the storage device 1220 and the processor 1202 via a high-speed I / O bus. In at least one embodiment, the I / O peripheral devices include, but are not limited to, an audio controller 1246, a network controller 1234, a firmware interface 1228, a wireless transceiver 1226, a touch sensor 1225, and a data storage device 1224 (e.g., a hard drive, flash memory, etc.). In at least one embodiment, the data storage device 1224 may be connected via a storage interface (e.g., SATA) or via a peripheral bus, such as a peripheral component interconnect bus (e.g., PCI, PCIe). In at least one embodiment, the touch sensor 1225 may include a touchscreen sensor, a pressure sensor, or a fingerprint sensor. In at least one embodiment, the wireless transceiver 1226 may be a Wi-Fi transceiver, a Bluetooth transceiver, or a mobile network transceiver, such as a 3G, 4G, or LTE transceiver. In at least one embodiment, the firmware interface 1228 enables communication with the system firmware and may be, for example, a Unified Extensible Firmware Interface (UEFI). In at least one embodiment, network controller 1234 may enable network connectivity to a wired network. In at least one embodiment, a high-performance network controller (not shown) is coupled to interface bus 1210. In at least one embodiment, audio controller 1246 is a multi-channel high-definition audio controller. In at least one embodiment, processing system 1200 includes an optional legacy I / O controller 1240 for coupling legacy (e.g., Personal System 2 (PS / 2)) devices to the system. In at least one embodiment, platform controller hub 1230 may also be connected to one or more Universal Serial Bus (USB) controllers 1242 that connect input devices, such as a keyboard and mouse combination 1243, a camera 1244, or other USB input devices.

[0099] In at least one embodiment, instances of the memory controller 1216 and platform controller hub 1230 may be integrated into a discrete external graphics processor, such as external graphics processor 1212. In at least one embodiment, the platform controller hub 1230 and / or memory controller 1216 may be external to one or more processors 1202. For example, in at least one embodiment, system 1200 may include external memory controller 1216 and platform controller hub 1230, which may be configured as a memory controller hub and peripheral controller hub in a system chipset communicating with processor 1202.

[0100] The inference and / or training logic 815 is used to perform inference and / or training operations associated with one or more embodiments. (The following is in conjunction with...) Figure 8A and / or Figure 8BDetails regarding the inference and / or training logic 815 are provided. In at least one embodiment, some or all of the inference and / or training logic 815 may be incorporated into the graphics processor 1500. For example, in at least one embodiment, the training and / or inference techniques described herein may use one or more ALUs embodied in the graphics processor. Furthermore, in at least one embodiment, the inference and / or training operations described herein may use, in addition to Figure 8A or Figure 8B The logic is performed using logic other than that shown. In at least one embodiment, the weight parameters may be stored in on-chip or off-chip memory and / or registers (shown or not shown), which configure the ALU of the graphics processor to execute one or more machine learning algorithms, neural network architectures, use cases, or training techniques described herein.

[0101] Such components can be used to support the execution of custom tasks using the same model (such as a large language model) without the need to retrain the model.

[0102] Figure 13 This is a block diagram of a processor 1300 having one or more processor cores 1302A-1302N, an integrated memory controller 1314, and an integrated graphics processor 1308 according to at least one embodiment. In at least one embodiment, the processor 1300 may include additional cores, up to and including additional cores 1302N, indicated by dashed boxes. In at least one embodiment, each processor core 1302A-1302N includes one or more internal cache units 1304A-1304N. In at least one embodiment, each processor core may also access one or more shared cache units 1306.

[0103] In at least one embodiment, internal cache units 1304A-1304N and shared cache unit 1306 represent a cache memory hierarchy within processor 1300. In at least one embodiment, cache memory units 1304A-1304N may include at least one level of instruction and data cache within each processor core and one or more levels of cache in a shared intermediate cache, such as Level 2 (L2), Level 3 (L3), Level 4 (L4), or other levels of cache, wherein the highest level of cache preceding external memory is classified as LLC. In at least one embodiment, cache coherence logic maintains coherence between the various cache units 1306 and 1304A-1304N.

[0104] In at least one embodiment, the processor 1300 may further include a set of one or more bus controller units 1316 and a system agent core 1310. In at least one embodiment, one or more bus controller units 1316 manage a set of peripheral buses, such as one or more PCI or PCIe buses. In at least one embodiment, the system agent core 1310 provides management functions for various processor components. In at least one embodiment, the system agent core 1310 includes one or more integrated memory controllers 1314 to manage access to various external memory devices (not shown).

[0105] In at least one embodiment, one or more processor cores 1302A-1302N include support for multi-threaded concurrent processing. In at least one embodiment, system agent core 1310 includes components for coordinating and operating cores 1302A-1302N during multi-threaded processing. In at least one embodiment, system agent core 1310 may additionally include a power control unit (PCU) including logic and components for regulating one or more power states of processor cores 1302A-1302N and graphics processor 1308.

[0106] In at least one embodiment, processor 1300 further includes a graphics processor 1308 for performing graph processing operations. In at least one embodiment, graphics processor 1308 is coupled to a shared cache unit 1306 and a system proxy core 1310 including one or more integrated memory controllers 1314. In at least one embodiment, system proxy core 1310 further includes a display controller 1311 for driving graphics processor outputs to one or more coupled displays. In at least one embodiment, display controller 1311 may also be a separate module coupled to graphics processor 1308 via at least one interconnect, or it may be integrated within graphics processor 1308.

[0107] In at least one embodiment, ring-based interconnect unit 1312 is used to couple internal components of processor 1300. In at least one embodiment, alternative interconnect units, such as point-to-point interconnects, switched interconnects, or other technologies, may be used. In at least one embodiment, graphics processor 1308 is coupled to ring interconnect 1312 via I / O link 1313.

[0108] In at least one embodiment, I / O link 1313 represents at least one of a variety of I / O interconnects, including packaged I / O interconnects that facilitate communication between various processor components and high-performance embedded memory module 1318 (e.g., eDRAM module). In at least one embodiment, each of processor cores 1302A-1302N and graphics processor 1308 uses embedded memory module 1318 as a shared last-level cache.

[0109] In at least one embodiment, processor cores 1302A-1302N are homogeneous cores executing a common instruction set architecture. In at least one embodiment, processor cores 1302A-1302N are heterogeneous in terms of instruction set architecture (ISA), with one or more processor cores 1302A-1302N executing a common instruction set, while one or more other processor cores 1302A-1302N execute a subset of the common instruction set or a different instruction set. In at least one embodiment, processor cores 1302A-1302N are heterogeneous in terms of microarchitecture, with one or more cores having relatively high power consumption coupled to one or more power cores having lower power consumption. In at least one embodiment, processor 1300 may be implemented on one or more chips or implemented as a SoC integrated circuit.

[0110] The inference and / or training logic 815 is used to perform inference and / or training operations associated with one or more embodiments. (The following is in conjunction with...) Figure 8A and / or Figure 8B Details regarding the inference and / or training logic 815 are provided. In at least one embodiment, some or all of the inference and / or training logic 815 may be incorporated into the processor 1300. For example, in at least one embodiment, the training and / or inference techniques described herein may use one or more ALUs embodied in... Figure 13 The graphics processor 1312, graphics core 1302A-1302N, or other components are used. Furthermore, in at least one embodiment, the inference and / or training operations described herein can use, except... Figure 8A or Figure 8B The logic is performed using logic other than that shown. In at least one embodiment, the weight parameters may be stored in on-chip or off-chip memory and / or registers (shown or not shown), which configure the ALU of the graphics processor 1300 to execute one or more machine learning algorithms, neural network architectures, use cases, or training techniques described herein.

[0111] Such components can be used to support the execution of custom tasks using the same model (such as a large language model) without the need to retrain the model.

[0112] Virtualization computing platform

[0113] Figure 14 This is an example data flow diagram of a process 1400 for generating and deploying an image processing and inference pipeline according to at least one embodiment. In at least one embodiment, process 1400 may be deployed for use with imaging devices, processing devices, and / or other device types at one or more facilities 1402. Process 1400 may be executed within a training system 1404 and / or a deployment system 1406. In at least one embodiment, training system 1404 may be used to train, deploy, and implement machine learning models (e.g., neural networks, object detection algorithms, computer vision algorithms, etc.) for use with deployment system 1406. In at least one embodiment, deployment system 1406 may be configured to offload processing and computing resources in a distributed computing environment to reduce the infrastructure requirements of facility 1402. In at least one embodiment, one or more applications in the pipeline may use or invoke services of deployment system 1406 (e.g., inference, visualization, computation, AI, etc.) during application execution.

[0114] In at least one embodiment, some applications used in the advanced processing and inference pipeline may use machine learning models or other AI to perform one or more processing steps. In at least one embodiment, a machine learning model may be trained at facility 1402 using data 1408 (e.g., imaging data) generated at facility 1402 (and stored on one or more Picture Archiving and Communication System (PACS) servers at facility 1402), imaging or sequencing data 1408 from another or more facilities, or a combination thereof. In at least one embodiment, training system 1404 may be used to provide applications, services, and / or other resources to generate a deployable machine learning model for the work of deploying system 1406.

[0115] In at least one embodiment, the model registry 1424 may be supported by an object storage system that supports version control and object metadata. In at least one embodiment, it may be available from within a cloud platform via, for example, cloud storage (e.g., Figure 12 The system uses a cloud-compatible application programming interface (API) to access object storage. In at least one embodiment, machine learning models within the model registry 1424 can be uploaded, listed, modified, or deleted by the developer or partner of the system interacting with the API. In at least one embodiment, the API can provide access to methods that allow users with appropriate credentials to associate models with applications, enabling the models to be executed as part of the containerized instantiation of the application.

[0116] In at least one embodiment, training pipeline 1404 ( Figure 14 This can include situations where facility 1402 is training its own machine learning model or has an existing machine learning model that needs optimization or updating. In at least one embodiment, imaging data 1408 generated by imaging devices, sequencing devices, and / or other types of devices can be received. In at least one embodiment, once the imaging data 1408 is received, AI-assisted annotation 1410 can be used to help generate annotations corresponding to the imaging data 1408 for use as ground-based data for the machine learning model. In at least one embodiment, AI-assisted annotation 1410 can include one or more machine learning models (e.g., convolutional neural networks (CNNs)) that can be trained to generate annotations corresponding to certain types of imaging data 1408 (e.g., from certain devices). In at least one embodiment, AI-assisted annotation 1410 can then be used directly or adjusted or fine-tuned using annotation tools to generate ground-based data. In at least one embodiment, AI-assisted annotation 1410, labeled clinical data 1412, or a combination thereof can be used as ground-based data for training the machine learning model. In at least one embodiment, the trained machine learning model may be referred to as output model 1416 and may be used by deployment system 1406 as described herein.

[0117] In at least one embodiment, training pipeline 1204 ( Figure 12This can include situations where facility 1402 requires a machine learning model to perform one or more processing tasks for deploying one or more applications in system 1406, but facility 1402 may not currently have such a machine learning model (or may not have an optimized, efficient, or effective model for this purpose). In at least one embodiment, an existing machine learning model can be selected from model registry 1424. In at least one embodiment, model registry 1424 can include machine learning models trained to perform various inference tasks on imaging data. In at least one embodiment, the machine learning model in model registry 1424 can be trained on imaging data from a different facility (e.g., a remote facility) instead of facility 1402. In at least one embodiment, the machine learning model may have already been trained on imaging data from one location, two locations, or any number of locations. In at least one embodiment, when training on imaging data from a specific location, training can be performed at that location, or at least in a manner that protects the confidentiality of the imaging data or restricts the transfer of the imaging data from off-site locations. In at least one embodiment, once a model has been trained or partially trained at a location, a machine learning model can be added to model registry 1424. In at least one embodiment, the machine learning model can then be retrained or updated at any number of other facilities, and the retrained or updated model can be used in model registry 1424. In at least one embodiment, a machine learning model (and referred to as output model 1416) can then be selected from model registry 1424, and can be executed in deployment system 1406 for one or more processing tasks for one or more applications of the deployment system.

[0118] In at least one embodiment, in training pipeline 1204 ( Figure 12In this scenario, the scenario may include facility 1402, which requires a machine learning model to perform one or more processing tasks for deploying one or more applications in system 1406, but facility 1402 may not currently have such a machine learning model (or may not have an optimized, efficient, or effective model). In at least one embodiment, the machine learning model selected from model registry 1424 may not be fine-tuned or optimized for the imaging data 1408 generated at facility 1402 due to population variability, robustness, anomalous diversity of training data, and / or other problems with the training data used to train the machine learning model. In at least one embodiment, AI-assisted annotation 1410 may be used to help generate annotations corresponding to the imaging data 1408 for use as ground-based data for training or updating the machine learning model. In at least one embodiment, labeled clinical data 1412 may be used as ground-based data for training the machine learning model. In at least one embodiment, retraining or updating the machine learning model may be referred to as model training 1414. In at least one embodiment, model training 1414 (e.g., AI-assisted annotation 1410, labeled clinical data 1412, or a combination thereof) can be used as ground-based real-world data to retrain or update the machine learning model. In at least one embodiment, the trained machine learning model can be referred to as output model 1416 and can be used by deployment system 1406, as described herein.

[0119] In at least one embodiment, deployment system 1406 may include software 1418, service 1420, hardware 1422, and / or other components, features, and functions. In at least one embodiment, deployment system 1406 may include a software "stack" such that software 1418 can be built on top of service 1420 and can be used to perform some or all of the processing tasks, and service 1420 and software 1418 can be built on top of hardware 1422 and use hardware 1422 to perform processing, storage, and / or other computational tasks of deployment system 1406. In at least one embodiment, software 1418 may include any number of different containers, each of which can perform an instantiation of an application. In at least one embodiment, each application can perform one or more processing tasks (e.g., inference, object detection, feature detection, segmentation, image enhancement, calibration, etc.) in a high-level processing and inference pipeline. In at least one embodiment, in addition to receiving and configuring imaging data for use by each container and / or by facility 1402 after processing through the pipeline, advanced processing and inference pipelines (e.g., to convert output back to available data types) can be defined based on the selection of different containers desired or required for processing imaging data 1408. In at least one embodiment, a combination of containers within software 1418 (e.g., constituting a pipeline) may be referred to as a virtual instrument (as described in more detail herein), and the virtual instrument may utilize service 1420 and hardware 1422 to perform some or all of the processing tasks of an application instantiated within the container.

[0120] In at least one embodiment, the data processing pipeline may receive input data (e.g., imaging data 1408) in a specific format in response to an inference request (e.g., a request from a user of deployment system 1406). In at least one embodiment, the input data may represent one or more images, videos, and / or other data representations generated by one or more imaging devices. In at least one embodiment, the data may be preprocessed as part of the data processing pipeline to prepare it for processing by one or more applications. In at least one embodiment, post-processing may be performed on the output of one or more inference tasks or other processing tasks of the pipeline to prepare output data for the next application and / or to prepare output data for user transmission and / or use (e.g., as a response to an inference request). In at least one embodiment, the inference task may be performed by one or more machine learning models, such as trained or deployed neural networks, which may include the output model 1416 of training system 1404.

[0121] In at least one embodiment, the tasks of the data processing pipeline can be encapsulated in containers, each container representing a discrete, fully functional instantiation of an application and a virtualized computing environment capable of referencing a machine learning model. In at least one embodiment, containers or applications can be published to a private (e.g., limited access) area of ​​a container registry (described in more detail herein), and trained or deployed models can be stored in model registry 1424 and associated with one or more applications. In at least one embodiment, an image of an application (e.g., a container image) can be used in the container registry, and once a user selects an image from the container registry for deployment in the pipeline, that image can be used to generate containers for instantiation of the application for use by the user's system.

[0122] In at least one embodiment, a developer (e.g., a software developer, clinician, physician, etc.) can develop, publish, and store an application (e.g., as a container) for performing image processing and / or inference on provided data. In at least one embodiment, a software development kit (SDK) associated with the system can be used to perform development, publication, and / or storage (e.g., to ensure that the developed application and / or container conforms to or is compatible with the system). In at least one embodiment, the developed application can be tested locally using the SDK (e.g., at a first facility, testing data from a first facility), the SDK serving as a system (e.g.,...). Figure 12 System 1200 may support at least some services 1420. In at least one embodiment, since DICOM objects may contain one to hundreds of images or other data types, and due to variations in the data, the developer may be responsible for managing (e.g., setting up constructs for preprocessing built into the application, etc.) the extraction and preparation of incoming data. In at least one embodiment, once verified by system 1400 (e.g., for accuracy), the application becomes available in the container registry for user selection and / or implementation to perform one or more processing tasks on data at the user's facility (e.g., a second facility).

[0123] In at least one embodiment, the developer can then share the application or container over a network for the system (e.g., Figure 14The system 1400 allows for user access and use. In at least one embodiment, completed and validated applications or containers may be stored in a container registry, and associated machine learning models may be stored in a model registry 1424. In at least one embodiment, a requesting entity (which provides an inference or image processing request) may browse the container registry and / or model registry 1424 to obtain applications, containers, datasets, machine learning models, etc., select desired combinations of elements to include in the data processing pipeline, and submit an image processing request. In at least one embodiment, the request may include input data necessary to execute the request (and, in some examples, patient-related data), and / or may include selections of applications and / or machine learning models to be executed when the request is processed. In at least one embodiment, the request may then be passed to one or more components of the deployment system 1406 (e.g., the cloud) to perform processing in the data processing pipeline. In at least one embodiment, processing performed by the deployment system 1406 may include referencing elements (e.g., applications, containers, models, etc.) selected from the container registry and / or model registry 1424. In at least one embodiment, once the results are generated through the pipeline, the results can be returned to the user for reference (e.g., for viewing in a suite of viewing applications executed locally, on a local workstation, or on a terminal).

[0124] In at least one embodiment, service 1420 may be utilized to assist in processing or executing applications or containers in the pipeline. In at least one embodiment, service 1420 may include computing services, artificial intelligence (AI) services, visualization services, and / or other service types. In at least one embodiment, service 1420 may provide functionality common to one or more applications in software 1418, thus abstracting functionality into services that can be invoked or utilized by applications. In at least one embodiment, the functionality provided by service 1420 can operate dynamically and more efficiently, while also allowing applications to process data in parallel (e.g., using...). Figure 12The parallel computing platform 1230 in the system can be scaled well. In at least one embodiment, it is not required that each application providing the same functionality as the shared service 1420 must have a corresponding instance of the service 1420, but the service 1420 can be shared between and among various applications. In at least one embodiment, as a non-limiting example, the service may include an inference server or engine that can be used to perform detection or segmentation tasks. In at least one embodiment, a model training service may be included, which can provide the ability to train and / or retrain machine learning models. In at least one embodiment, a data augmentation service may be further included, which can provide GPU-accelerated data (e.g., DICOM, RIS, CIS, conforming to REST, RPC, raw, etc.) extraction, resizing, scaling, and / or other enhancements. In at least one embodiment, a visualization service may be used, which can add image rendering effects (e.g., ray tracing, rasterization, denoising, sharpening, etc.) to add realism to two-dimensional (2D) and / or three-dimensional (3D) models. In at least one embodiment, a virtual instrument service may be included, which provides beamforming, segmentation, inference, imaging, and / or support for other applications within the virtual instrument pipeline.

[0125] In at least one embodiment, where service 1420 includes an AI service (e.g., an inference service), as part of application execution, one or more machine learning models can be executed by invoking (e.g., as an API call) the inference service (e.g., an inference server) to execute one or more machine learning models or their processing. In at least one embodiment, where another application includes one or more machine learning models for a segmentation task, the application can invoke the inference service to execute the machine learning models for performing one or more processing operations associated with the segmentation task. In at least one embodiment, software 1418 implementing advanced processing and inference pipelines, including a segmentation application and an anomaly detection application, can be pipelined because each application can invoke the same inference service to execute one or more inference tasks.

[0126] In at least one embodiment, hardware 1422 may include a GPU, CPU, graphics card, AI / deep learning system (e.g., an AI supercomputer, such as NVIDIA's DGX), cloud platform, or a combination thereof. In at least one embodiment, different types of hardware 1422 may be used to provide efficient, specially built support for software 1418 and services 1420 in deployment system 1406. In at least one embodiment, GPU processing may be used to perform local processing (e.g., at facility 1402) within the AI / deep learning system, in the cloud system, and / or other processing components of deployment system 1406 to improve the efficiency, accuracy, and performance of image processing and generation. In at least one embodiment, as a non-limiting example, software 1418 and / or services 1420 may be optimized for GPU processing in relation to deep learning, machine learning, and / or high-performance computing. In at least one embodiment, at least some of the computing environment of deployment system 1406 and / or training system 1404 may be executed in a data center, one or more supercomputers, or high-performance computing systems with GPU-optimized software (e.g., a hardware and software combination of an NVIDIA DGX system). In at least one embodiment, as described herein, hardware 1422 may include any number of GPUs that can be invoked to perform data processing in parallel. In at least one embodiment, the cloud platform may also include GPU-optimized execution for deep learning tasks, GPU processing for machine learning tasks, or other computational tasks. In at least one embodiment, an AI / deep learning supercomputer and / or GPU-optimized software (e.g., as provided on NVIDIA's DGX systems) may be used as a hardware abstraction and scaling platform to execute the cloud platform (e.g., NVIDIA's NGC). In at least one embodiment, the cloud platform may integrate application container cluster systems or coordination systems (e.g., Kubernetes) across multiple GPUs to achieve seamless scaling and load balancing.

[0127] Figure 15 This is a system diagram of an example system 1500 for generating and deploying an imaging deployment pipeline according to at least one embodiment. In at least one embodiment, system 1500 can be used to implement... Figure 14 The process 1400 and / or other processes include advanced processing and inference pipelines. In at least one embodiment, system 1500 may include training system 1404 and deployment system 1406. In at least one embodiment, training system 1404 and deployment system 1406 may be implemented using software 1418, service 1420 and / or hardware 1422, as described herein.

[0128] In at least one embodiment, system 1500 (e.g., training system 1404 and / or deployment system 1406) may be implemented in a cloud computing environment (e.g., using cloud 1526). In at least one embodiment, system 1500 may be implemented locally (in relation to a healthcare facility) or as a combination of cloud computing resources and local computing resources. In at least one embodiment, access to the API in cloud 1526 may be restricted to authorized users by establishing security measures or protocols. In at least one embodiment, the security protocol may include a network token, which may be signed by an authentication service (e.g., AuthN, AuthZ, Gluecon, etc.) and may carry appropriate authorization. In at least one embodiment, the API of the virtual instrument (described herein) or other instances of system 1500 may be restricted to a set of public IPs that have been audited or authorized for interaction.

[0129] In at least one embodiment, the various components of system 1500 may communicate with each other using any of a variety of different network types, including but not limited to local area networks (LANs) and / or wide area networks (WANs) via wired and / or wireless communication protocols. In at least one embodiment, communication between facilities and components of system 1500 (e.g., for sending inference requests, for receiving the results of inference requests, etc.) may be transmitted via one or more data buses, wireless data protocols (Wi-Fi), wired data protocols (e.g., Ethernet), etc.

[0130] In at least one embodiment, similar to the description herein. Figure 14 As described, training system 1404 can execute training pipeline 1504. In at least one embodiment, where deployment system 1406 uses one or more machine learning models in deployment pipeline 1510, training pipeline 1504 can be used to train or retrain one or more (e.g., pre-trained) models, and / or implement one or more pre-trained models 1506 (e.g., without retraining or updating). In at least one embodiment, as a result of training pipeline 1504, output model 1416 can be generated. In at least one embodiment, training pipeline 1504 can include any number of processing steps, such as, but not limited to, transformation or adaptation of imaging data (or other input data). In at least one embodiment, different training pipelines 1504 can be used for different machine learning models used by deployment system 1406. In at least one embodiment, similar to the description of... Figure 14 The training pipeline 1504 described in the first example can be used for the first machine learning model, similar to the one described above. Figure 14 The training pipeline 1504 described in the second example can be used for a second machine learning model, similar to the one described above. Figure 14The training pipeline 1504 of the third example described can be used for a third machine learning model. In at least one embodiment, any combination of tasks within the training system 1404 can be used according to the requirements of each corresponding machine learning model. In at least one embodiment, one or more machine learning models may have already been trained and are ready for deployment, so the training system 1404 may not perform any processing on the machine learning models, and one or more machine learning models may be implemented by the deployment system 1406.

[0131] In at least one embodiment, depending on the implementation or embodiment, the output model 1416 and / or the pre-trained model 1506 may include any type of machine learning model. In at least one embodiment, and not limited thereto, the machine learning model used by system 1500 may include models using linear regression, logistic regression, decision trees, support vector machines (SVM), Naive Bayes, k-nearest neighbors (Knn), k-means clustering, random forests, dimensionality reduction algorithms, gradient boosting algorithms, neural networks (e.g., autoencoders, convolutions, recursion, perceptrons, long / short-term memory (LSTM), Hopfield, Boltzmann, deep belief, deconvolution, generative adversarial, liquid state machines, etc.), and / or other types of machine learning models.

[0132] In at least one embodiment, the training pipeline 1504 may include AI-assisted annotations, as described herein regarding at least Figure 15B. More detailed description. In at least one embodiment, labeled clinical data 1412 can be generated using any number of techniques (e.g., conventional annotation). In at least one embodiment, in some examples, labels or other annotations can be generated by drawing programs (e.g., annotation programs), computer-aided design (CAD) programs, tagging programs, another type of application suitable for generating annotations or labels for ground reality, and / or can be hand-drawn. In at least one embodiment, ground reality data can be synthetically generated (e.g., generated from computer models or renderings), realistically generated (e.g., designed and generated from real-world data), automatically generated by machines (e.g., extracting features from data using feature analysis and learning, and then generating labels), manually annotated (e.g., taggers or annotation experts, defining the placement of labels), and / or combinations thereof. In at least one embodiment, for each instance of imaging data 1408 (or other data types used by machine learning models), there may be corresponding ground reality data generated by training system 1404. In at least one embodiment, AI-assisted annotation can be performed as part of deployment pipeline 1510; supplementing or replacing AI-assisted annotation included in training pipeline 1504. In at least one embodiment, system 1500 may include a multi-layer platform, which may include a software layer (e.g., software 1418) of a diagnostic application (or other application type) capable of performing one or more medical imaging and diagnostic functions. In at least one embodiment, system 1500 may be communicatively coupled (e.g., via an encrypted link) to a network of PACS servers of one or more facilities. In at least one embodiment, system 1500 may be configured to access and reference data from PACS servers to perform operations such as training machine learning models, deploying machine learning models, image processing, inference, and / or other operations.

[0133] In at least one embodiment, the software layer may be implemented as a secure, encrypted, and / or certified API that can invoke (e.g., call) an application or container from an external environment (e.g., facility 1402). In at least one embodiment, the application may then invoke or execute one or more services 1420 to perform computational, AI, or visualization tasks associated with their respective applications, and the software 1418 and / or service 1420 may utilize the hardware 1422 to perform processing tasks efficiently and effectively.

[0134] In at least one embodiment, deployment system 1406 may execute deployment pipeline 1510. In at least one embodiment, deployment pipeline 1510 may include any number of applications, which may be sequential, non-sequential, or otherwise applied to imaging data (and / or other data types) – including AI-assisted annotation, the imaging data being generated by imaging devices, sequencing devices, genomics devices, etc., as described above. In at least one embodiment, as described herein, deployment pipeline 1510 for an individual device may be referred to as a virtual instrument for the device (e.g., a virtual ultrasound instrument, a virtual CT scanner, a virtual sequencing instrument, etc.). In at least one embodiment, for a single device, more than one deployment pipeline 1510 may exist, depending on the desired information from the data generated from the device. In at least one embodiment, a first deployment pipeline 1510 may exist if it is desired to detect an anomaly from an MRI machine, and a second deployment pipeline 1510 may exist if it is desired to perform image enhancement from the output of the MRI machine.

[0135] In at least one embodiment, the image generation application may include processing tasks that utilize machine learning models. In at least one embodiment, a user may wish to use their own machine learning model or select a machine learning model from the model registry 1424. In at least one embodiment, a user may implement their own machine learning model or select a machine learning model to be included in the application performing the processing tasks. In at least one embodiment, the application may be optional and customizable, and by defining the application's construction, the deployment and implementation of the application for a specific user is presented as a more seamless user experience. In at least one embodiment, by leveraging other features of system 1500 (e.g., service 1420 and hardware 1422), the deployment pipeline 1510 can be more user-friendly, provide easier integration, and produce more accurate, efficient, and timely results.

[0136] In at least one embodiment, deployment system 1406 may include user interface 1514 (e.g., graphical user interface, web interface, etc.) which may be used to select applications to be included in deployment pipeline 1510, deploy applications, modify or change applications or their parameters or configurations, use and interact with deployment pipeline 1510 during setup and / or deployment, and / or otherwise interact with deployment system 1406. In at least one embodiment, although not shown with respect to training system 1404, user interface 1514 (or different user interfaces) may be used to select models to be used in deployment system 1406, to select models to be trained or retrained in training system 1404, and / or to otherwise interact with training system 1404.

[0137] In at least one embodiment, in addition to the application coordination system 1528, a pipeline manager 1512 may also be used to manage interactions between the application or container deploying pipeline 1510 and services 1420 and / or hardware 1422. In at least one embodiment, the pipeline manager 1512 may be configured to facilitate interactions from application to application, from application to service 1420, and / or from application or service to hardware 1422. In at least one embodiment, although shown as included in software 1418, this is not intended to be limiting, and in some examples (e.g., as...) Figure 12 As shown in the diagram (cc), pipeline manager 1512 may be included in service 1420. In at least one embodiment, application coordination system 1528 (e.g., Kubernetes, DOCKER, etc.) may include container coordination system that can group applications into containers as logical units for coordination, management, scaling, and deployment. In at least one embodiment, by associating applications (e.g., rebuilding applications, splitting applications, etc.) from deployment pipeline 1510 with individual containers, each application can execute in a self-contained environment (e.g., at the kernel level) to improve speed and efficiency.

[0138] In at least one embodiment, each application and / or container (or its image) can be developed, modified, and deployed independently (e.g., a first user or developer can develop, modify, and deploy a first application, and a second user or developer can develop, modify, and deploy a second application separate from the first user or developer). This allows focus on the tasks of a single application and / or container without being hindered by the tasks of another application or container. In at least one embodiment, the pipeline manager 1512 and the application coordination system 1528 can facilitate communication and collaboration between different containers or applications. In at least one embodiment, the application coordination system 1528 and / or the pipeline manager 1512 can facilitate communication and resource sharing between and within each application or container, provided that the expected inputs and / or outputs of each container or application are known to the system (e.g., based on the application or container's architecture). In at least one embodiment, since one or more applications or containers in the deployment pipeline 1510 can share the same services and resources, the application coordination system 1528 can coordinate, load balance, and determine the sharing of services or resources between and within the various applications or containers. In at least one embodiment, the scheduler can be used to track the resource requirements of applications or containers, the current or planned use of these resources, and resource availability. Therefore, in at least one embodiment, the scheduler can allocate resources to different applications and distribute resources between and among applications, taking into account the system's needs and availability. In some examples, the scheduler (and / or other components of the application coordination system 1528) can determine resource availability and distribution based on constraints imposed on the system (e.g., user constraints), such as Quality of Service (QoS), the urgency of data output (e.g., to determine whether to perform real-time processing or delayed processing), etc.

[0139] In at least one embodiment, service 1420, utilized and shared by applications or containers in deployment system 1406, may include computing service 1516, AI service 1518, visualization service 1520, and / or other service types. In at least one embodiment, an application may invoke (e.g., execute) one or more services 1420 to perform processing operations for the application. In at least one embodiment, an application may utilize computing service 1516 to perform supercomputing or other high-performance computing (HPC) tasks. In at least one embodiment, one or more computing services 1516 may be utilized to perform parallel processing (e.g., using parallel computing platform 1530) to process data substantially simultaneously through one or more applications and / or one or more tasks of a single application. In at least one embodiment, parallel computing platform 1530 (e.g., NVIDIA's CUDA) may implement general-purpose computing on a GPU (GPGPU) (e.g., GPU 1522). In at least one embodiment, the software layer of parallel computing platform 1530 may provide access to the GPU's virtual instruction set and parallel computing elements to execute computing kernels. In at least one embodiment, the parallel computing platform 1530 may include memory, and in some embodiments, memory may be shared between and within multiple containers, and / or between and within different processing tasks within a single container. In at least one embodiment, inter-process communication (IPC) calls may be generated for multiple containers and / or multiple processes within containers to enable the use of the same data (e.g., multiple different stages of one or more applications processing the same information) from a shared memory segment of the parallel computing platform 1530. In at least one embodiment, instead of copying data and moving it to different locations in memory (e.g., read / write operations), the same data in the same memory location can be used for any number of processing tasks (e.g., at the same time, at different times, etc.). In at least one embodiment, this information about the new location of the data can be stored and shared between applications because the resulting data from processing is used to generate new data. In at least one embodiment, the location of the data, and the location of the updated or modified data, may be part of the definition of how the payload in the container is understood.

[0140] In at least one embodiment, AI service 1518 may be used to perform an inference service for executing a machine learning model associated with an application (e.g., a task to perform one or more processing tasks of the application). In at least one embodiment, AI service 1518 may utilize AI system 1524 to execute a machine learning model (e.g., a neural network such as a CNN) for segmentation, reconstruction, object detection, feature detection, classification, and / or other inference tasks.

[0141] In at least one embodiment, the application deploying pipeline 1510 can use one or more output models 1416 of the self-training system 1404 and / or other models of the application to perform inference on imaging data. In at least one embodiment, two or more examples of inference using the application coordination system 1528 (e.g., a scheduler) may be available. In at least one embodiment, a first category may include high-priority / low-latency paths that can implement higher service level protocols, such as for performing inference on urgent requests in emergency situations or for radiologists during diagnostic procedures. In at least one embodiment, a second category may include standard priority paths that can be used for requests that may not be urgent or for situations where analysis can be performed at a later time. In at least one embodiment, the application coordination system 1528 may allocate resources (e.g., services 1420 and / or hardware 1422) based on priority paths for different inference tasks of the AI ​​service 1518.

[0142] In at least one embodiment, shared memory may be installed into AI service 1518 in system 1500. In at least one embodiment, shared memory may operate as a cache (or other storage device type) and may be used to process inference requests from applications. In at least one embodiment, when an inference request is submitted, a set of API instances of deployment system 1406 may receive the request and may select one or more instances (e.g., for best fit, for load balancing, etc.) to process the request. In at least one embodiment, to process the request, the request may be fed into a database, and if not already in the cache, a machine learning model may be located from model registry 1424. A verification step may ensure that an appropriate machine learning model is loaded into the cache (e.g., shared memory), and / or a copy of the model may be saved to the cache. In at least one embodiment, if the application is not already running or there are not enough instances of the application, a scheduler (e.g., the scheduler of pipeline manager 1512) may be used to start the application referenced in the request. In at least one embodiment, if an inference server has not yet been started to execute the model, an inference server may be started. Any number of inference servers may be started for each model. In at least one embodiment, in a pull model that clusters inference servers, the model can be cached whenever load balancing is favorable. In at least one embodiment, the inference servers can be statically loaded into the corresponding distributed servers.

[0143] In at least one embodiment, an inference server running in a container can be used to perform inference. In at least one embodiment, an instance of the inference server can be associated with a model (and optionally multiple versions of the model). In at least one embodiment, if an instance of the inference server does not exist when a request to perform inference on the model is received, a new instance can be loaded. In at least one embodiment, when the inference server is started, a model can be passed to the inference server, allowing the same container to be used to serve different models, as long as the inference server runs as different instances.

[0144] In at least one embodiment, during application execution, an inference request for a given application can be received, and a container (e.g., an instance of a hosted inference server) can be loaded (if not already loaded), and a launcher can be invoked. In at least one embodiment, preprocessing logic within the container can (e.g., using a CPU and / or GPU) load, decode, and / or perform any additional preprocessing on the incoming data. In at least one embodiment, once the data is ready for inference, the container can infer the data as needed. In at least one embodiment, this can include a single inference call for an image (e.g., a hand X-ray) or can request inference for hundreds of images (e.g., a chest CT scan). In at least one embodiment, the application can summarize the results before completion, which may include, but is not limited to, a single confidence score, pixel-level segmentation, voxel-level segmentation, generating visualizations, or generating text to summarize the results. In at least one embodiment, different priorities can be assigned to different models or applications. For example, some models may have a real-time (TAT less than 1 minute) priority, while other models may have a lower priority (e.g., TAT less than 10 minutes). In at least one embodiment, model execution time can be measured from the requesting agency or entity, and may include cooperative network traversal time and inference service execution time.

[0145] In at least one embodiment, the transfer of requests between service 1420 and the inference application can be hidden behind a software development kit (SDK) and robust transfer can be provided via queues. In at least one embodiment, requests are placed in queues via an API for individual application / tenant ID combinations, and the SDK pulls requests from the queues and provides them to the application. In at least one embodiment, the name of the queue can be provided in the environment where the SDK picks up the queue. In at least one embodiment, asynchronous communication via queues may be useful because it allows any instance of the application to pick up work when it becomes available. Results can be sent back via queues to ensure no data loss. In at least one embodiment, queues can also provide the ability to partition work, as the highest priority work can go into a queue connected to a majority of instances of the application, while the lowest priority work can go into a queue connected to a single instance that processes tasks in the order they are received. In at least one embodiment, the application can run on a GPU-accelerated instance generated in cloud 1526, and the inference service can perform inference on the GPU.

[0146] In at least one embodiment, visualization service 1520 can be used to generate visualizations for viewing the output of application and / or deployment pipeline 1510. In at least one embodiment, visualization service 1520 can utilize GPU 1522 to generate visualizations. In at least one embodiment, visualization service 1520 can implement rendering effects such as ray tracing to generate higher quality visualizations. In at least one embodiment, visualizations can include, but are not limited to, 2D image rendering, 3D volume rendering, 3D volume reconstruction, 2D tomographic slicing, virtual reality display, augmented reality display, etc. In at least one embodiment, a virtualized environment can be used to generate virtual interactive displays or environments (e.g., virtual environments) for system users (e.g., doctors, nurses, radiologists, etc.) to interact with. In at least one embodiment, visualization service 1520 can include an internal visualizer, cinematic and / or other rendering or image processing capabilities or functions (e.g., ray tracing, rasterization, internal optics, etc.).

[0147] In at least one embodiment, hardware 1422 may include GPU 1522, AI system 1524, cloud 1526, and / or any other hardware for performing training system 1404 and / or deployment system 1406. In at least one embodiment, GPU 1522 (e.g., NVIDIA's TESLA and / or QUADRO GPUs) may include any number of GPUs that can be used to perform processing tasks of any feature or function of computing service 1516, AI service 1518, visualization service 1520, other services, and / or software 1418. For example, for AI service 1518, GPU 1522 may be used to perform preprocessing on imaging data (or other data types used by machine learning models), postprocessing on the output of machine learning models, and / or perform inference (e.g., to execute machine learning models). In at least one embodiment, cloud 1526, AI system 1524, and / or other components of system 1500 may use GPU 1522. In at least one embodiment, cloud 1526 may include a GPU-optimized platform for deep learning tasks. In at least one embodiment, AI system 1524 may use a GPU, and one or more AI systems 1524 may be used to perform cloud 1526 (or at least part of a task for deep learning or inference). Similarly, although hardware 1422 is shown as a discrete component, this is not intended to be limiting, and any component of hardware 1422 may be combined with or utilized by any other component of hardware 1422.

[0148] In at least one embodiment, AI system 1524 may include a specially built computing system (e.g., a supercomputer or HPC) configured for inference, deep learning, machine learning, and / or other artificial intelligence tasks. In at least one embodiment, in addition to CPU, RAM, memory, and / or other components, features, or functions, AI system 1524 (e.g., NVIDIA's DGX) may also include GPU-optimized software (e.g., a software stack) that can be executed using multiple GPUs 1522. In at least one embodiment, one or more AI systems 1524 may be implemented in a cloud 1526 (e.g., in a data center) to perform some or all of the AI-based processing tasks of system 1500.

[0149] In at least one embodiment, cloud 1526 may include GPU-accelerated infrastructure (e.g., NVIDIA's NGC) that can provide a GPU-optimized platform for performing processing tasks of system 1500. In at least one embodiment, cloud 1526 may include AI system 1524 for performing one or more AI-based tasks of system 1500 (e.g., as a hardware abstraction and scaling platform). In at least one embodiment, cloud 1526 may be integrated with application coordination system 1528 utilizing multiple GPUs to achieve seamless scaling and load balancing between and within applications and services 1420. In at least one embodiment, as described herein, cloud 1526 may be responsible for performing at least some of the services 1420 of system 1500, including computing service 1516, AI service 1518, and / or visualization service 1520. In at least one embodiment, cloud 1526 may perform large and small batch inference (e.g., perform NVIDIA's TENSORRT), provide accelerated parallel computing APIs and platform 1530 (e.g., NVIDIA's CUDA), perform application coordination system 1528 (e.g., KUBERNETES), provide graphics rendering APIs and platform (e.g., for ray tracing, 2D graphics, 3D graphics and / or other rendering techniques to produce higher quality cinematic effects), and / or provide other functionalities for system 1500.

[0150] Figure 15 A illustrates a data flow diagram of a process 1500 for training, retraining, or updating a machine learning model according to at least one embodiment. In at least one embodiment, a non-limiting example may be used. Figure 15 The system 1500 executes the process 1500. In at least one embodiment, the process 1500 may utilize the services 1420 and / or hardware 1422 of the system 1500, as described herein. In at least one embodiment, the refined model 1512 generated by the process 1500 may be executed by the deployment system 1406 for one or more containerized applications in the deployment pipeline 1510.

[0151] In at least one embodiment, model training 1414 may include retraining or updating the initial model 1504 (e.g., a pre-trained model) using new training data (e.g., new input data, such as customer dataset 1506, and / or new ground reality data associated with the input data). In at least one embodiment, to retrain or update the initial model 1504, the output or loss layer of the initial model 1504 may be reset or deleted, and / or replaced with an updated or new output or loss layer. In at least one embodiment, the initial model 1504 may have previously finely tuned parameters (e.g., weights and / or biases) retained from previous training, so training or retraining 1414 may not require as much time or processing as training the model from scratch. In at least one embodiment, during model training 1414, by resetting or replacing the output or loss layer of the initial model 1504, on a new customer dataset 1506 (e.g., new input data, such as customer dataset 1506, and / or new ground reality data associated with the input data), the initial model 1504 may be retrained or updated. Figure 14 When generating predictions on image data (1408), the parameters of the new dataset can be updated and readjusted based on the loss calculation associated with the accuracy of the output or loss layer.

[0152] In at least one embodiment, the pre-trained model 1506 may be stored in a data storage or registry (e.g., Figure 14 (Model registry 1424). In at least one embodiment, the pre-trained model 1506 may have been trained at least partially at one or more facilities other than the facility performing process 1500. In at least one embodiment, to protect the privacy and rights of patients, subjects, or customers at different facilities, the pre-trained model 1506 may have been trained locally using locally generated customer or patient data. In at least one embodiment, the pre-trained model 1506 may be trained using cloud 1526 and / or other hardware 1422, but confidential, privacy-protected patient data may not be transferred to, used by, or accessed by any component of cloud 1526 (or other non-local hardware). In at least one embodiment, if the pre-trained model 1506 is trained using patient data from more than one facility, the pre-trained model 1506 may have been trained separately for each facility before training on patient or customer data from another facility. In at least one embodiment, such as when customer or patient data has been published for privacy reasons (e.g., by abandonment, for experimental purposes, etc.), or where customer or patient data is included in a public dataset, customer or patient data from any number of facilities can be used to train a pre-trained model 1506 locally and / or externally, such as in a data center or other cloud computing infrastructure.

[0153] In at least one embodiment, when selecting an application for use in deployment pipeline 1510, the user may also select a machine learning model for a specific application. In at least one embodiment, the user may not have a model available, so the user may select a pre-trained model 1506 to use with the application. In at least one embodiment, the pre-trained model 1506 may not be optimized to generate accurate results on the user facility's customer dataset 1506 (e.g., based on patient diversity, demographics, type of medical imaging equipment used, etc.). In at least one embodiment, the pre-trained model 1506 may be updated, retrained, and / or fine-tuned for use at various facilities before being deployed to deployment pipeline 1510 for use with one or more applications.

[0154] In at least one embodiment, a user may select a pre-trained model 1506 to be updated, retrained, and / or fine-tuned, and the pre-trained model 1506 may be referred to as the initial model 1504 of the training system 1404 in process 1500. In at least one embodiment, a client dataset 1506 (e.g., imaging data, genomic data, sequencing data, or other data types generated by equipment at the facility) may be used to perform model training 1414 (which may include, but is not limited to, transfer learning) on ​​the initial model 1504 to generate a refined model 1512. In at least one embodiment, ground-based data corresponding to the client dataset 1506 may be generated by the training system 1404. In at least one embodiment, ground-based data (e.g., such as...) may be generated at the facility at least in part by clinicians, scientists, physicians, practitioners, etc. Figure 14 Clinical data marked in 1412).

[0155] In at least one embodiment, AI-assisted annotation 1410 may be used in some examples to generate ground reality data. In at least one embodiment, AI-assisted annotation 1410 (e.g., implemented using an AI-assisted annotation SDK) may leverage machine learning models (e.g., neural networks) to generate ground reality data for suggestions or predictions for a customer dataset. In at least one embodiment, user 1510 may use the annotation tool within a user interface (graphical user interface (GUI)) on computing device 1508.

[0156] In at least one embodiment, user 1510 can interact with the GUI via computing device 1508 to edit or fine-tune annotations or automatic annotations. In at least one embodiment, polygon editing features can be used to move the vertices of a polygon to more precise or fine-tuned positions.

[0157] In at least one embodiment, once the customer dataset 1506 has associated ground-based data, the ground-based data (e.g., from AI-assisted annotations, manual labeling, etc.) can be used to generate a refined model 1512 during model training 1414. In at least one embodiment, the customer dataset 1506 can be applied to the initial model 1504 an arbitrary number of times, and the ground-based data can be used to update the parameters of the initial model 1504 until an acceptable level of accuracy is achieved for the refined model 1512. In at least one embodiment, once the refined model 1512 is generated, it can be deployed within one or more deployment pipelines 1510 at the facility to perform one or more processing tasks related to medical imaging data.

[0158] In at least one embodiment, the refined model 1512 can be uploaded to the pre-trained model 1506 in the model registry 1424 for selection by another facility. In at least one embodiment, this process can be completed at any number of facilities, allowing the refined model 1512 to be further refined any number of times on a new dataset to generate a more general model.

[0159] Figure 15 B is an example illustration of a client-server architecture 1532 for enhancing an annotation tool using a pre-trained annotation model, according to at least one embodiment. In at least one embodiment, an AI-assisted annotation tool 1536 may be instantiated based on the client-server architecture 1532. In at least one embodiment, the annotation tool 1536 in an imaging application can assist ray surgeons, for example, in identifying organs and abnormalities. In at least one embodiment, the imaging application may include software tools, as a non-limiting example, that help user 1510 identify several extreme points on a specific organ of interest in a raw image 1534 (e.g., in a 3D MRI or CT scan) and receive automatic annotation results for all 2D slices of the specific organ. In at least one embodiment, the results may be stored in a data store as training data 1538 and used as (e.g., but not limited to) ground-based real-world data for training. In at least one embodiment, when computing device 1508 sends extreme points for AI-assisted annotation 1410, for example, a deep learning model may receive this data as input and return inference results for segmenting organs or abnormalities. In at least one embodiment, a pre-instantiated annotation tool (e.g., Figure 15The AI-assisted annotation tool 1536B in B can be enhanced by making API calls (e.g., API call 1544) to a server (such as annotation assistant server 1540), which may include a set of pre-trained models 1542 stored, for example, in an annotation model registry. In at least one embodiment, the annotation model registry may store pre-trained models 1542 (e.g., machine learning models, such as deep learning models) that have been pre-trained to perform AI-assisted annotation on specific organs or abnormalities. In at least one embodiment, these models can be further updated using a training pipeline 1504. In at least one embodiment, the pre-installed annotation tool can be improved over time as newly labeled clinical data 1412 is added.

[0160] Such components can be used to support the execution of custom tasks using the same model (such as a large language model) without the need to retrain the model.

[0161] Figure 16A A data flow diagram of a process 1600 for training, retraining, or updating a machine learning model according to at least one embodiment is shown. In at least one embodiment, a non-limiting example can be used. Figure 15 System 1500 executes process 1600. In at least one embodiment, process 1600 may utilize the services and / or hardware described herein. In at least one embodiment, the refined model 1612 generated by process 1600 may be executed by a deployment system of one or more containerized applications in a deployment pipeline.

[0162] In at least one embodiment, model training 1614 may include retraining or updating the initial model 1604 (e.g., a pre-trained model) using new training data (e.g., new input data, such as customer dataset 1606, and / or new ground reality data associated with the input data). In at least one embodiment, to retrain or update the initial model 1604, the output or loss layer of the initial model 1604 may be reset, deleted, and / or replaced with an updated or new output or loss layer. In at least one embodiment, the initial model 1604 may have previously fine-tuned parameters (e.g., weights and / or biases) retained from previous training, so training or retraining 1614 may not require as much time or processing as training the model from scratch. In at least one embodiment, during model training 1614, by resetting or replacing the output or loss layer of the initial model 1604, when generating predictions on the new customer dataset 1606, the parameters may be updated and readjusted for the new dataset based on loss calculations associated with the accuracy of the output or loss layer.

[0163] In at least one embodiment, the pre-trained model 1606 may be stored in a data store or registry. In at least one embodiment, the pre-trained model 1606 may be trained at least partially at one or more facilities other than the facility performing process 1600. In at least one embodiment, to protect the privacy and rights of patients, subjects, or customers at different facilities, the pre-trained model 1606 may have been trained locally using locally generated customer or patient data. In at least one embodiment, the pre-trained model 1406 may be trained using the cloud and / or other hardware, but confidential, privacy-protected patient data may not be transferred to any component of the cloud (or other non-local hardware), or used or accessed by it. In at least one embodiment, if the pre-trained model 1606 is trained using patient data from more than one facility, the pre-trained model 1606 may have been trained separately for each facility before being trained on patient or customer data from another facility. In at least one embodiment, such as when customer or patient data has been published with privacy concerns (e.g., by waiver, for experimental use, etc.), or when customer or patient data is included in a public dataset, customer or patient data from any number of facilities can be used to train a pre-trained model 1606 locally and / or externally, such as in a data center or other cloud computing infrastructure.

[0164] In at least one embodiment, when selecting an application for the deployment pipeline, the user may also select a machine learning model for a specific application. In at least one embodiment, the user may not have a model to use, so the user may select a pre-trained model to use with the application. In at least one embodiment, the pre-trained model may not have been optimized to produce accurate results on the user's facility's customer dataset 1606 (e.g., based on patient diversity, demographics, type of medical imaging equipment used, etc.). In at least one embodiment, the pre-trained model may be updated, retrained, and / or fine-tuned for use at the appropriate facility before being deployed into the deployment pipeline for use with one or more applications.

[0165] In at least one embodiment, the user can select a pre-trained model to update, retrain, and / or fine-tune, and this pre-trained model may be referred to as the initial model 1604 of the training system within process 1600. In at least one embodiment, a client dataset 1606 (e.g., imaging data, genomics data, sequencing data, or other data types generated by equipment at the facility) can be used to perform model training (which may include, but is not limited to, transfer learning) on ​​the initial model 1604 to produce a refined model 1612. In at least one embodiment, ground-based data corresponding to the client dataset 1606 may be generated by the training system 1304. In at least one embodiment, the ground-based data may be generated at the facility, at least in part, by clinicians, scientists, physicians, or practitioners.

[0166] In at least one embodiment, AI-assisted annotation may be used to generate ground-based data in some examples. In at least one embodiment, AI-assisted annotation (e.g., implemented using an AI-assisted annotation SDK) may leverage machine learning models (e.g., neural networks) to generate suggested or predicted ground-based data for a customer dataset. In at least one embodiment, a user may use the annotation tool within a user interface (graphical user interface (GUI)) on a computing device.

[0167] In at least one embodiment, user 1610 can interact with the GUI via computing device 1608 to edit or fine-tune (automatic) annotations. In at least one embodiment, polygon editing features can be used to move the vertices of a polygon to more precise or fine-tuned positions.

[0168] In at least one embodiment, once the client dataset 1606 has associated ground-based data, the ground-based data (e.g., from AI-assisted annotations, manual labeling, etc.) can be used to generate a refined model 1612 during model training. In at least one embodiment, the client dataset 1606 can be applied to the initial model 1604 any number of times, and the ground-based data can be used to update the parameters of the initial model 1604 until an acceptable level of accuracy is achieved for the refined model 1612. In at least one embodiment, once the refined model 1612 is generated, it can be deployed in one or more deployment pipelines at the facility to perform one or more processing tasks related to medical imaging data.

[0169] In at least one embodiment, the refined model 1612 can be uploaded to a model registry of pre-trained models for selection by another facility. In at least one embodiment, this process can be performed at any number of facilities, allowing the refined model 1612 to be further refined an arbitrary number of times on new datasets to generate a more general model.

[0170] Figure 16B This is an example illustration of a client-server architecture 1632 for enhancing an annotation tool using a pre-trained annotation model, according to at least one embodiment. In at least one embodiment, an AI-assisted annotation tool 1636 may be instantiated based on the client-server architecture 1632. In at least one embodiment, the annotation tool 1636 in an imaging application can assist radiologists, for example, in identifying organs and abnormalities. In at least one embodiment, the imaging application may include software tools, as a non-limiting example, that help user 1610 identify several extreme points on a specific organ of interest in a raw image 1634 (e.g., in a 3D MRI or CT scan) and receive automatic annotation results for all 2D slices of that specific organ. In at least one embodiment, the results may be stored in a data store as training data 1638 and used as (e.g., but not limited to) ground-based real-world data for training. In at least one embodiment, when computing device 1608 sends extreme points for AI-assisted annotation, for example, a deep learning model may receive this data as input and return inference results for segmenting organs or abnormalities. In at least one embodiment, a pre-instantiated annotation tool (e.g., Figure 16B The AI-assisted annotation tool 1636B can be enhanced by making API calls (e.g., API call 1644) to a server (e.g., annotation assistant server 1640), which may include a set of pre-trained models 1642 stored, for example, in an annotation model registry. In at least one embodiment, the annotation model registry may store pre-trained models 1642 (e.g., machine learning models, such as deep learning models) that have been pre-trained to perform AI-assisted annotation on specific organs or abnormalities. These models can be further updated using a training pipeline. In at least one embodiment, the pre-installed annotation tool can be improved over time as new labeled data is added.

[0171] Other variations are within the spirit of this disclosure. Therefore, while the disclosed technology is readily adaptable to various modifications and alternative constructions, certain embodiments thereof are illustrated in the accompanying drawings and have been described in detail above. However, it should be understood that the disclosure is not intended to be limited to one or more specific forms disclosed, but rather, it is intended to cover all modifications, alternative constructions, and equivalents falling within the spirit and scope of this disclosure as defined in the appended claims.

[0172] Unless otherwise stated or obviously contradicted by the context, the terms “a,” “an,” and “the,” and similar references, used in the context of describing the disclosed embodiments (particularly in the context of the appended claims), should be interpreted as encompassing both singular and plural forms, rather than as definitions of the terms. Unless otherwise stated, the terms “comprising,” “having,” “including,” and “containing” should be interpreted as open-ended terms (meaning “including, but not limited to”). The term “connection” (referring to a physical connection where not modified) should be interpreted as partially or wholly contained, attached to, or joined together, even with some intervention. Unless otherwise indicated herein, references to numerical ranges herein are intended only as a way of abbreviating each individual value falling within that range, and each individual value is incorporated into the specification as if it were separately described herein. Unless otherwise indicated or contradicted by the context, the use of the terms “set” (e.g., “item set”) or “subset” should be interpreted as a non-empty set comprising one or more members. Furthermore, unless otherwise indicated or contradicted by the context, the term "subset" of a corresponding set does not necessarily refer to an appropriate subset of the corresponding set, but rather the subset and the corresponding set can be equal.

[0173] Unless otherwise explicitly stated or clearly contradicted by the context, connective phrases such as “at least one of A, B, and C” or “at least one of A, B, and C” are understood in the context to generally refer to items, terms, etc., which can be A or B or C, or any non-empty subset of the set A, B, and C. For example, in an illustrative example of a set with three members, the connective phrases “at least one of A, B, and C” and “at least one of A, B, and C” refer to any of the following sets: {A}, {B}, {C}, {A, B}, {A, C}, {B, C}, {A, B, C}. Therefore, such connective language is generally not intended to imply that some embodiments require the presence of at least one of A, at least one of B, and at least one of C. Additionally, unless otherwise stated or contradicted by the context, the term “multiple” indicates a plural state (e.g., “multiple items” means multiple items). The number of items in a multiple item is at least two, but may be more if explicitly indicated or indicated by the context. Furthermore, unless otherwise stated or clearly understood from the context, the phrase “based on” means “at least partially based on” rather than “based on only”.

[0174] Unless otherwise indicated herein or clearly contradicted by the context, the operations of the processes described herein may be performed in any suitable order. In at least one embodiment, processes such as those described herein (or variations thereof and / or combinations thereof) are executed under the control of one or more computer systems configured with executable instructions and are implemented as code (e.g., executable instructions, one or more computer programs, or one or more application programs) that is executed jointly on one or more processors via hardware or a combination thereof. In at least one embodiment, the code is stored on a computer-readable storage medium, for example, in the form of a computer program comprising a plurality of instructions executable by one or more processors. In at least one embodiment, the computer-readable storage medium is a non-transitory computer-readable storage medium that excludes transient signals (e.g., propagating transient electrical or electromagnetic transmissions) but includes non-transitory data storage circuitry (e.g., buffers, caches, and queues). In at least one embodiment, code (e.g., executable code or source code) is stored on one or more non-transitory computer-readable storage media (or other memory for storing executable instructions) on which executable instructions are stored, which, when executed by one or more processors of a computer system (i.e., as a result of execution), cause the computer system to perform the operations described herein. In at least one embodiment, the set of non-transitory computer-readable storage media comprises multiple non-transitory computer-readable storage media, and one or more of the individual non-transitory storage media lack all the code, but the multiple non-transitory computer-readable storage media collectively store all the code. In at least one embodiment, the executable instructions are executed such that different instructions are executed by different processors; for example, the non-transitory computer-readable storage media store the instructions, and the main central processing unit (“CPU”) executes some instructions while the graphics processing unit (“GPU”) executes other instructions. In at least one embodiment, different components of the computer system have separate processors, and the different processors execute different subsets of the instructions.

[0175] Therefore, in at least one embodiment, the computer system is configured to implement one or more services that perform the operations of the processes described herein, either individually or collectively, and such a computer system is configured with suitable hardware and / or software to enable the implementation of the operations. Furthermore, the computer system implementing at least one embodiment of this disclosure is a single device, and in another embodiment it is a distributed computer system comprising multiple devices operating in different ways, such that the distributed computer system performs the operations described herein, and that a single device does not perform all the operations.

[0176] The use of any and all examples or exemplary language (e.g., “such as”) provided herein is intended only to better illustrate embodiments of this disclosure and does not constitute a limitation on the scope of the disclosure unless otherwise required. No language in the specification should be construed as indicating that any unclaimed element is essential to the practice of the disclosure.

[0177] All references cited in this article, including publications, patent applications and patents, are incorporated herein by reference as if each reference were individually and specifically indicated to be incorporated herein by reference and the entire contents of which are described herein.

[0178] The terms “coupled” and “connected”, and their derivatives, may be used in the specification and claims. It should be understood that these terms may not be intended to be synonyms with each other. Rather, in certain examples, “connected” or “coupled” may be used to indicate that two or more elements are in direct or indirect physical or electrical contact with each other. “Coupled” may also mean that two or more elements are not in direct contact with each other, but still cooperate or interact with each other.

[0179] Unless otherwise expressly stated, it will be understood that throughout this specification, terms such as “processing,” “computing,” “determining,” etc., refer to the actions and / or processes of a computer or computing system or similar electronic computing device that process and / or convert data represented as physical quantities (e.g., electrons) in the registers and / or memory of the computing system into other data represented as physical quantities in the memory, registers, or other such information storage, transmission, or display devices of the computing system.

[0180] In a similar manner, the term "processor" can refer to any device or part of memory that processes electronic data from registers and / or memory and converts that electronic data into other electronic data that can be stored in registers and / or memory. As a non-limiting example, a "processor" can be a CPU or a GPU. A "computing platform" can include one or more processors. As used herein, a "software" process can include, for example, software and / or hardware entities that perform work over time, such as tasks, threads, and intelligent agents. Similarly, each process can refer to multiple processes that execute instructions sequentially or intermittently, sequentially, or in parallel. The terms "system" and "method" are used interchangeably herein, provided that a system can embody one or more methods, and a method can be considered a system.

[0181] This document refers to the process of acquiring, obtaining, receiving, or inputting analog or digital data into a subsystem, computer system, or computer-implemented machine. Analog and digital data can be acquired, obtained, received, or input in various ways, such as by receiving data as a parameter to a function call or a call to an application programming interface (API). In some implementations, the process of acquiring, obtaining, receiving, or inputting analog or digital data can be accomplished by transmitting data via a serial or parallel interface. In another implementation, the process of acquiring, obtaining, receiving, or inputting analog or digital data can be accomplished by transmitting data from a providing entity to an acquiring entity via a computer network. Reference can also be made to providing, outputting, transmitting, sending, or presenting analog or digital data. In various examples, the process of providing, outputting, transmitting, sending, or presenting analog or digital data can be implemented by transmitting data as an input or output parameter to a function call, an API, or an inter-process communication mechanism.

[0182] While the discussion above illustrates example implementations of the described technologies, other architectures can be used to implement the described functionality and are intended to fall within the scope of this disclosure. Furthermore, although specific assignments of responsibilities have been defined above for discussion purposes, various functions and responsibilities can be assigned and divided in different ways depending on the circumstances.

[0183] Furthermore, although the subject matter has been described in language specific to structural features and / or methodological actions, it should be understood that the subject matter claimed in the appended claims is not necessarily limited to the specific features or actions described. Rather, specific features and actions are disclosed as exemplary forms for implementing the claims.

Claims

1. A method comprising: Determine the natural language text string associated with the request based at least on a request received at one of the endpoints of a system of two or more communicatively coupled computing devices; Based at least on the natural language text string and the task associated with the endpoint, determine one or more guidance mechanisms associated with the request; The natural language text string is processed using a large language model and based at least on one or more of the guidance mechanisms to generate a text result corresponding to the request, wherein the large language model has not been trained to perform at least a subset of one or more tasks associated with at least one of the plurality of endpoints; as well as Using the endpoint, a response to the request is generated, the response including at least the text result.

2. The method of claim 1, wherein each of the plurality of endpoints is associated with one or more corresponding tasks.

3. The method of claim 2, further comprising: For each of the plurality of endpoints, select one or more guidance mechanisms from a corresponding group for the corresponding task of each endpoint.

4. The method of claim 1, wherein the one or more guidance mechanisms include a prompt token indicating at least one of a reasoning type to be performed for the task or a result type to be returned for the task.

5. The method of claim 1, wherein the one or more guidance mechanisms include retrieval set labels indicating one or more datasets to be referenced, wherein the results are further generated based at least on processing data retrieved from the one or more datasets at least based on the retrieval set labels using the language model.

6. The method of claim 1, wherein the one or more guidance mechanisms include adapter weights for modifying at least one of the network weights or layer structures of the language model prior to the processing.

7. The method of claim 1, further comprising: Generate one or more alphanumeric strings representing the one or more guidance mechanisms; as well as The one or more alphanumeric strings are added to the beginning of the natural language text string to form a modified text string. Processing the natural language text string includes processing the modified text string.

8. The method of claim 1, further comprising: Using one or more of the multiple endpoints, generate multiple text strings for multiple tasks that will be performed using the large language model; as well as The multiple text strings are transmitted as at least one of the following: a batch or more batches, or combinations of homogeneous task flows.

9. The method of claim 1, wherein the large language model is associated with two or more model instances of different sizes, and wherein the endpoint is trained against a specific model instance among the two or more model instances to perform the task.

10. The method of claim 1, wherein the natural language text string is obtained from the request according to one or more orchestration rules for configuring the endpoint.

11. At least one processor, comprising: One or more circuits are configured to cause the at least one processor to perform operations, the operations including: Determine the natural language text string associated with the request based at least on a request received at one of the endpoints of a system of two or more communicatively coupled computing devices; Based at least on the natural language text string and the task associated with the endpoint, determine one or more guidance mechanisms associated with the request; The natural language text string is processed using a large language model and based at least on one or more of the guidance mechanisms to generate a text result corresponding to the request, wherein the large language model has not been trained to perform at least a subset of one or more tasks associated with at least one of the plurality of endpoints; and Using the endpoint, a response to the request is generated, the response including at least the text result.

12. The at least one processor of claim 11, wherein each of the plurality of endpoints is associated with one or more corresponding tasks.

13. The at least one processor of claim 12, wherein the operation further comprises: Select one or more guiding mechanisms for each of the plurality of endpoints for the corresponding task of each endpoint.

14. The at least one processor of claim 12, wherein the operation further comprises: Generate one or more alphanumeric strings representing the one or more guidance mechanisms; as well as The one or more alphanumeric strings are added to the beginning of the natural language text string to form a modified text string. Processing the natural language text string includes processing the modified text string.

15. The at least one processor of claim 11, wherein the large language model is associated with two or more model instances of different sizes, and wherein the endpoint is trained to perform the task for a specific model instance among the two or more model instances.

16. The at least one processor as claimed in claim 11, wherein the at least one processor is comprised of at least one of: A system used to perform simulation operations; A system used to perform simulations to test or validate autonomous machine applications; A system used for rendering graphics output; A system used to perform deep learning operations; Systems implemented using edge devices; Systems used to generate or present virtual reality (VR) content; A system for generating or presenting augmented reality (AR) content; A system for generating or presenting mixed reality (MR) content; A system containing one or more virtual machines (VMs); A system that is at least partially implemented in a data center; A system for using simulation to perform hardware testing; Systems for generating synthetic data; A collaborative content creation platform for 3D assets; or A system that utilizes cloud computing resources at least in part.

17. A system comprising: One or more processors are configured to generate a response to the request based on a response format associated with an endpoint corresponding to the request, the response being generated at least based on reasoning performed by a large language model according to one or more guiding mechanisms determined to be associated with the response format. The large language model was not trained to perform at least a subset of one or more tasks associated with the endpoint.

18. The system of claim 17, wherein the one or more guidance mechanisms include at least one of cue tokens, retrieval tag sets, or adapter weights.

19. The system of claim 17, wherein the one or more guidance mechanisms are used to perform at least one of the following: Change one or more weights in at least one layer of the language model. Change the structure of one or more layers of the language model. Update the input of the language model corresponding to the request, or Provides an instruction to access the dataset to retrieve data corresponding to the input of the language model.

20. The system of claim 17, wherein the system comprises at least one of the following: A system used to perform simulation operations; A system used to perform simulations to test or validate autonomous machine applications; A system used for rendering graphics output; A system used to perform deep learning operations; Systems implemented using edge devices; Systems used to generate or present virtual reality (VR) content; A system for generating or presenting augmented reality (AR) content; A system for generating or presenting mixed reality (MR) content; A system containing one or more virtual machines (VMs); A system that is at least partially implemented in a data center; A system for using simulation to perform hardware testing; Systems for generating synthetic data; A collaborative content creation platform for 3D assets; or A system that utilizes cloud computing resources at least in part.

Images