EP4681112A1

EP4681112A1 - Machine learning based agent for text editing

Info

Publication number: EP4681112A1
Application number: EP24726453.4A
Authority: EP
Inventors: Lei MENG; Lei Shu; Yinxiao Liu; Tony Mak; Chang Li; Yun Zhu; Abhanshu Sharma; Matthew Sharifi; Hen Fitoussi; Jayakumar Hoskere; Simon Tong
Original assignee: Google LLC
Current assignee: Google LLC
Priority date: 2023-04-19
Filing date: 2024-04-19
Publication date: 2026-01-21
Also published as: WO2024220845A1; CN121153038A

Abstract

Provided is a computing system that includes a machine learning agent that can be requested to perform various editing tasks in, for example, a word processing application or other text-generation environments. The machine learning agent can include one or more machine-learned language models that have been trained on a set of training data that provides examples of edits performed on long-form text.

Description

MACHINE LEARNING BASED AGENT FOR TEXT EDITING

FIELD

[0001] The present disclosure relates generally to a machine learning agent for performing text editing. More particularly, the present disclosure relates to a computing system that includes a machine learning agent that can be requested to perform various editing tasks in, for example, a word processing application or other text-generation environments. The machine learning agent can include one or more machine-learned language models that have been trained on a set of training data that provides examples of edits performed on long-form text.

BACKGROUND

[0002] Various approaches have been suggested for using machine learning models to perform editing of text. Many of these proposed approaches are limited to performing discrete tasks on sentence-level text. The sentence-level models are bounded by the information in a single sentence without its surrounding context, and thus may result in suggestions that are not coherent with its surrounding context, leading to low semantic and logical consistency on the paragraph or document level. Further, these sentence-level models typically do not perform well when the input texts are longer than a few sentences. Other approaches have suggested using so-called "large language models” or “LLMs” to perform document-level editing. However, general purpose LLMs are typically very large (e.g., in terms of number of parameters and required computation) and therefore incur significant computational costs, often rendering them undesirable to use for production.

SUMMARY

[0003] Aspects and advantages of embodiments of the present disclosure will be set forth in part in the following description, or can be learned from the description, or can be learned through practice of the embodiments.

[0004] One example aspect is directed to a computing system for automated text editing. The computing system includes a machine learning agent comprising one or more machine- learned models configured to perform one or more text editing tasks. The computing system includes a text generation application, wherein the machine learning agent comprises an addressable entity that is addressable within the text generation application. The computing system is configured to perform operations, the operations comprising: obtaining, by the computing system, data descriptive of an annotation input by a user in the text generation application, wherein the annotation corresponds to a set of textual content created within the text generation application, and wherein the annotation comprises a natural language text editing request that is addressed to the machine learning agent; in response to the annotation, executing, by the computing system, the machine learning agent to cause the machine learning agent to generate one or more edits to the set of textual content based on the natural language text editing request; and updating, by the computing system, the set of textual content based on the one or more edits generated by the machine learning agent.

[0005] In some implementations, the annotation comprises a comment displayed in a margin of a graphical user interface of the text generation application. In some implementations, the annotation is addressed to the machine learning agent using natural language. In some implementations, the annotation is addressed to the machine learning agent using a particular symbol that designates the machine learning agent a recipient of the annotation. In some implementations, the set of textual content comprises two or more sentences. In some implementations, the set of textual content comprises an entirety of a document being edited by the user in the text generation application. In some implementations, the set of textual content comprises a subset of a document being edited by the user in the text generation application, and wherein the set of textual content has been selected by the user. In some implementations, executing, by the computing system, the machine learning agent comprises: generating, by the computing system, a model prompt based on the set of textual content and the natural language text editing request; and processing, by the computing system, the model prompt with the one or more machine- learned models of the machine learning agent to generate the one or more edits as an output of the one or more machine-learned models. In some implementations, generating, by the computing system, the model prompt based on the set of textual content and the natural language text editing request comprises: combining, by the computing system, (1) the set of textual content, (2) one or more context portions that are adjacent to the set of textual content within a document being edited by the user in the text generation application, and (3) the natural language text editing request to form the model prompt. In some implementations, the operations further comprise: executing, by the computing system, the machine learning agent to cause the machine learning agent to generate a second annotation that is responsive to the annotation input by the user; and inserting, by the computing sy stem, the second annotation into the text generation application. [0006] Another example aspect is directed to a computer-implemented method to train a language model. The method includes obtaining, by a computing system comprising one or more computing devices, a training tuple comprising a natural language text editing instruction, a set of textual content, and data describing one or more ground-truth edits to the set of textual content performed in associated with the natural language text editing instruction, wherein the set of textual content comprises two or more sentences. The method includes processing, by the computing system, the natural language text editing instruction and the set of textual content with the language model to generate one or more predicted edits to the set of textual content that are responsive to the natural language text editing instruction. The method includes evaluating, by the computing system, a loss function that generates a loss value based on a comparison of the one or more predicted edits to the one or more ground-truth edits. The method includes modifying, by the computing system, one or more parameter values of the language model based on the loss function.

[0007] In some implementations, the training tuple comprises a comment and edit history obtained from a text generation application in which the language model is deployed. In some implementations, the training tuple comprises an edit history obtained from a large text edit corpus. In some implementations, the large text edit corpus comprises edits to an online collaboratively-edited encyclopedia. In some implementations, the natural language instruction comprises a summary- of the one or more ground-truth edits. In some implementations, the natural language instruction comprises a modified version of a summary of the one or more ground-truth edits, wherein the modified version of the summary was produced by a second, different language model from an original version of the summan- of the one or more ground-truth edits. In some implementations, the natural language instruction has been selected by a reward language model. In some implementations, the natural language instruction has been filtered based on an natural language inference (NLI) score, length, or an edit distance. The NLI score can indicate an inference relation between two (e.g., ordered) texts such as pre- and post-editing texts. The inference relation can indicate, for example, entailment, contradiction, or neutral. In some implementations, the natural language instruction compnses a synthetic prompt.

[0008] Other aspects of the present disclosure are directed to various systems, apparatuses, non-transitory computer-readable media, user interfaces, and electronic devices. [0009] These and other features, aspects, and advantages of various embodiments of the present disclosure will become better understood with reference to the following description and appended claims. The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate example embodiments of the present disclosure and, together with the description, serve to explain the related principles.

BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Detailed discussion of embodiments directed to one of ordinary skill in the art is set forth in the specification, which makes reference to the appended figures, in which:

[0011] Figure 1 depicts a block diagram of an example computing system according to example embodiments of the present disclosure.

[0012] Figures 2A-I depicts a sequence of example graphical user interfaces demonstrating example interactions with an example machine learning agent according to example embodiments of the present disclosure.

[0013] Figures 3A-B depict block diagrams of example inference processes for a machine-learned language model according to example embodiments of the present disclosure.

[0014] Figures 4A-B depict block diagrams of example training processes for a machine- learned language model according to example embodiments of the present disclosure.

[0015] Figure 5 is a flow chart diagram illustrating an example method for training a machine-learned model according to example implementations of aspects of the present disclosure;

[0016] Figure 6 is a block diagram of an example processing flow for using machine- learned model(s) to process input(s) to generate output(s) according to example implementations of aspects of the present disclosure;

[0017] Figure 7 is a block diagram of an example sequence processing model according to example implementations of aspects of the present disclosure;

[0018] Figure 8 is a block diagram of an example technique for populating an example input sequence for processing by a sequence processing model according to example implementations of aspects of the present disclosure;

[0019] Figure 9 is a block diagram of an example model development platform according to example implementations of aspects of the present disclosure;

[0020] Figure 10 is a block diagram of an example training workflow for training a machine-learned model according to example implementations of aspects of the present disclosure; [0021] Figure 11 is a block diagram of an inference system for operating one or more machine-learned model(s) to perform inference according to example implementations of aspects of the present disclosure;

[0022] Figure 12 is a block diagram of an example networked computing system according to example implementations of aspects of the present disclosure;

[0023] Figure 13 is a block diagram of an example computing device according to example implementations of aspects of the present disclosure; and

[0024] Figure 14 is a block diagram of an example computing device according to example implementations of aspects of the present disclosure.

[0025] Reference numerals that are repeated across plural figures are intended to identify the same features in various implementations.

DETAILED DESCRIPTION

Overview

[0026] Generally, the present disclosure is directed to a machine learning agent included in or provided as a feature of a text generation application such as a web-based collaborative word processing application or other productivity tool that enables the generation of text. The machine learning agent can be called upon to perform various text editing tasks, including document-level content co-creation tasks. In particular, according to one aspect of the present disclosure, the machine learning agent can be treated as an addressable entity within the word processing application. As an addressable entity, the machine learning agent can be summoned or requested to provide edits to a set of textual content created within the word processing application.

[0027] As one example, a user can add a comment or other annotation to the set of textual content. The user can address the comment to the machine learning agent (e.g., by explicitly marking the comment as directed toward the agent or otherwise addressing the agent within the comment). The user-generated comment can also include a request for editing expressed in a natural language. In response, the machine learning agent can edit the set of textual content as requested by the comment. In some cases, the machine learning agent can also respond to the user’s original comment with its own responsive comment (e g., which may be an explanation of the edits performed). In such fashion, a user can call the machine learning agent into action via a comment or other annotation and the machine learning agent can respond by performing the requested editing operations. [0028] Another aspect of the present disclosure is directed to techniques for training the machine learning agent to perform the editing tasks described herein. In particular, the machine learning agent can include one or more machine-learned language models. The language models can be trained on a set of training data that provides examples of edits performed on long-form text. The term “long-form text'’ can refer to sets of textual content that include two or more sentences. The training data can include examples of pre- and postedit text and can also include corresponding natural language editing instructions that describe (e.g., request) the example edits. Additionally or alternatively, the training data can include natural language comments (e.g., explanations of the edits) that are responsive to the original natural language editing instructions. The language models can be trained (e.g., finetuned) on this training data to enable the models to perform long-form text editing in response to a user-generated edit request.

[0029] The training data can be created from a number of different sources. As one example, the training data can be created from annotation and edit histories from the same word processing application or other productivity tool in which the machine learning agent is to be deployed. As another example, the training data can be created from edit histories from large text corpuses, such as editing histories for an online, collaboratively-edited encyclopedia. For each edit in the edit history⁷ of the text corpus, a summary⁷ of the edit may be available. This summary can be included in the training dataset as an example editing instruction. The summary of the edit can be included as-is or can be modified or refined via application of human changes or edits made by a separate language model. More generally, a number of pre- or post- processing techniques can also be performed on the training data to improve the performance of the language model trained on the training data. Thus, training data can be generated which enables one or more language models to perform generalized text editing of long-form textual content.

[0030] The systems and methods of the present disclosure provide a number of technical effects and benefits. As one example, the present disclosure provides a general purpose editing model that is not constrained to specific tasks, but instead is able to perform any number of different types of editing actions observed in a training dataset. As another example technical effect, the present disclosure reduces training costs and the number of labeled data needed to develop features in the content co-creation domain. In particular, the proposed model can be quickly adapted to new features and generate training data for any downstream tasks. Reduced training costs can correspond to reduced usage of computational resources such as processor usage, memory usage, and/or network bandwidth usage. As yet another example technical effect, the present disclosure extends state-of-the-art LLM technology to long-form editing tasks such as document-level editing tasks.

[0031] With reference now to the Figures, example embodiments of the present disclosure will be discussed in further detail.

Example Systems for Automated Text Editing

[0032] Figure 1 depicts a block diagram of an example computing system 100 according to example embodiments of the present disclosure. The computing system 100 includes a text generation application 112. The text generation application 112 can be a software program or application that allows users to create, edit, format, and manage textual content (e.g., stored or contained in a file or document) on a computer. The text generation application 112 can provide tools and features for creating, editing, and formatting text, as well as options for saving, printing, and sharing documents.

[0033] The text generation application 112 can be any number of different applications that enable one or more users to create textual content.

[0034] In one example, the text generation application 112 can be a word processing application or other text editor. The word processing application can be used for creating and editing text documents and can provide a wide range of features for creating and formatting text content, including text editing, formatting, styling, and layout options.

[0035] In another example, the text generation application 112 can be a blogging platform or a social media platform. These platforms can allow users to create text content in the form of posts, tweets, updates, and articles.

[0036] In another example, the text generation application 112 can be an electronic mail ("‘email”) client. The email client can allow users to create text content in the form of emails. [0037] In another example, the text generation application 112 can be a platform for generating and editing programming language content. The application 1 12 can be used by programmers, developers, and other professionals for creating and editing computer code, scripts, or other programming language content.

[0038] In another example, the text generation application 112 can be a note-taking application. The note-taking application can allow users to create and organize text content in the form of notes, lists, and memos.

[0039] In yet further examples, the text generation application 112 can be or include a spreadsheet software application, a presentation software application, task or project management tools, time management tools, or cloud storage and file sharing tools. [0040] The text generation application 112 can receive one or more user inputs 114 that result in the creation or editing of a set of text content 116. The inputs 114 can come from one user or multiple different users. The inputs 114 can be received via user interaction with a graphical user interface. As an example, in some implementations, the text generation application 112 can be a web-based collaborative word processing application that enables one or more users to interact and collaborate, through the Internet, to create or edit the set of textual content 116 in real-time, regardless of their physical location.

[0041] Thus, user(s) can provide input(s) 1 14 to make changes to the set of textual content, including changes to the text, formatting, and other elements using the editing tools provided by the text generation application 112. These changes can include adding, deleting, or modifying text, formatting text (e.g., changing font styles, sizes, colors), and adding or modifying other elements such as images, tables, or hyperlinks.

[0042] The text generation application 112 can also enable the users to generate comments 117 or other annotations for or with respect to the set of textual content 116. For example, users can leave comments 117 on the set of textual content 116 to provide feedback, suggestions, or clarification. Comments 117 can appear as annotations or notes in the margins of the document, or as pop-up boxes, and/or can be viewed by other users with access to the document. Users can respond to comments, resolve comments, or edit comments as needed. [0043] According to an aspect of the present disclosure, the computing system 100 can also include a machine learning agent 118. The machine learning agent 118 can be or include a computer program or system that uses machine learning algorithms and techniques to perform tasks or make decisions autonomously or semi-autonomously, based on input data and feedback from its environment. The machine learning agent 118 can be designed to leam from experience, adapt to changing conditions, and optimize its performance over time without being explicitly programmed for specific tasks. Although the machine learning agent 118 is shown in Figure 1 as separate from the text generation application 112, in other cases the machine learning agent 118 may be included in or be a feature of the text generation application 112. In other cases, the machine learning agent 118 can exchange data with the text generation application 112 using an application programming interface (API).

[0044] The machine learning agent 118 can interact with the text generation application 112. For example, the machine learning agent 118 can receive data descriptive of the set of textual content 116 and/or any annotations associated therewith.

[0045] The machine learning agent 118 can include and implement one or more machine-learned language models 120. In some implementations, the machine-learned language model 120 can be or can be referred to as a so-called “large language model’'. In other implementations, the machine-learned language model 120 can be or can be referred to as a so-called “large multimodal model” (LMM). Large language models refer to powerful machine learning models that are capable of processing and generating text on a large scale. These models are ty pically trained (e.g., “pre-trained”) on massive amounts of data, often millions or even billions of sentences, to leam the patterns, structures, and nuances of natural languages. Large multimodal models include language models that can also process one or more additional modalities of data beyond language or text data. For example, the additional modalities of data can be image data (e.g., including video data), audio data, sensor data, time-series data, and/or other modalities of data. In implementations in which a model (e.g., a language model) processes a sequence of inputs and/or generates a sequence of outputs, such model can be referred to as a “sequence processing model”.

[0046] The language model(s) 120 can be trained using various techniques or architectures, such as recurrent neural networks (RNNs), transformers, or other machine learning algorithms. Training can occur in two stages: (1) pre-training on a large unlabeled text corpus, and then (2) fine tuning on a specific task such as a short- and/or long-form editing task. Language models 120 can be trained to handle a wide range of natural language processing (NLP) tasks, such as text generation, text completion, sentiment analysis, machine translation, question answering, and more. Examples of large language models include BERT (Bidirectional Encoder Representations from Transformers) and T5 (Text-to-Text Transfer Transformer). Additional examples include the Gemini family of LMMs.

[0047] According to an aspect of the present disclosure, the machine learning agent 118 can be called upon to perform various text editing tasks, including document-level content co-creation tasks. In particular, according to one aspect of the present disclosure, the machine learning agent 118 can be treated as an addressable entity within the text generation application 112. As an addressable entity, the machine learning agent 118 can be summoned or requested to provide edits to the set of textual content 116 created within the text generation application 112.

[0048] As described above, a user can add a comment 117 or other annotation to the set of textual content 116. As an example, in the context of the text generation application 1 12, a comment can refer to a t pe of annotation or note that can be added to the set of textual content 116 to provide additional information, feedback, or suggestions without altering the set of textual content 116 itself. Comments 117 can be used for collaboration purposes, allowing multiple users to provide input and feedback on a document. They can be visible in the margins or as pop-up boxes within the document, and can be viewed, edited, and deleted as needed. Comments 117 can identify the person who added the comment, along with a timestamp, to identify the commenter and provide context.

[0049] According to an aspect of the present disclosure, a user can address a comment 117 to the machine learning agent 118. As one example, the user can explicitly mark the comment 117 as directed toward the machine learning agent 118. As one example, the user may mark the comment 117 as directed the machine learning agent 118 (or use some other symbol to formally designate a recipient of the comment). The "@" feature, commonly known as the "at" feature or the "mention" feature can allow users to mention or tag other users (or the agent 118) by using the symbol followed by the username or handle of the person they want to notify or reference.

[0050] In another example, the user can informally address the comment 117 to the machine learning agent 118. Filters implemented using computer logic can detect that the machine learning agent 118 is an intended recipient of the comment.

[0051] The user-generated comment 117 can include a request for editing expressed in a natural language. Natural language refers to the language used by humans for communication, including spoken and written forms of expression. The request for editing can generally convey the need for text editing or can more specifically describe desired areas of improvement such as flow, quality, readability, effectiveness, coherence, grammar, punctuation, conciseness, engagement, and/or positivity. The natural language request for editing can also request a certain tone or describe the intended audience.

[0052] In response to the comment 117 that is addressed to the machine learning agent 118, the machine learning agent 118 can edit the set of textual content 116 as requested by the comment 117. For example, the machine learning agent 118 can process the comment 117 and the set of textual content 116 with the one or more language models 120 to generate edited text. In some implementations, the agent 118 can interact with the text generation application 112 to directly replace the existing set of textual content 116 with the edited content generated by the model 120. In other cases, the user may be given the opportunity to review the edited content and select whether or not to replace the existing set of textual content 116 with the edited content generated by the model 120.

[0053] In some implementations, the edited content generated by the model 120 can be a full set of replacement text. In some implementations, the edited content generated by the model 120 can describe specific edits to the existing set of textual content 116. In some implementations, the edited content generated by the model 120 can be provided in a clean format (without edits or changes explicitly shown). In some implementations, the edited content generated by the model 120 can be provided in a markup format (e g., with edits or changes shown using visual cues such as strikethrough, underlining, highlighting, and/or font color, etc.).

[0054] In some cases, the machine learning agent 118 can also respond to the user’s original comment 117 with its own responsive comment 117 (e g., which may be an explanation of the edits performed). For example, the new comment 117 generated by the machine learning agent 118 can be provided in a comment chain or other comment history associated w ith the set of textual content 116.

[0055] Thus, a user can call the machine learning agent 118 into action via a comment 117 or other annotation and the machine learning agent 118 can respond by performing the requested editing operations. As an example of this process. Figures 2A-I depicts a sequence of example graphical user interfaces demonstrating example interactions with an example machine learning agent according to example embodiments of the present disclosure.

[0056] As shown in Figure 2A, a text generation application has received user input that has generated an initial set of textual content (“We’re organizing a summit in Zurich. Zurich is a great place to visit ”)

[0057] As shown in Figure 2B, the user has highlighted a certain portion of the set of textual content (“Zurich is a great place to visit”). The user has used the graphical user interface to create a new comment.

[0058] As shown in Figure 2C, the user has addressed the comment to the machine learning agent (in this example the agent is referred to using the example name of “CoComposer”).

[0059] As shown in Figure 2D, the user has entered a natural language description of the requested edits (“Could you expand on this a bit?”).

[0060] As shown in Figure 2E, in response to the user’s comment, the machine learning agent has edited the document to add additional textual content (e.g., content that describes Zurich). The machine learning agent has also provided an additional comment in the comment thread (“I expanded it. Please take a look.”)

[0061] As shown in Figure 2F, the user and the machine learning agent have both added further comments responsive to each other.

[0062] As shown in Figure 2G, the user has added additional textual content (“I hope the summit will be nice.”) [0063] As shown in Figure 2H, the user has again addressed a comment to the machine learning agent with an additional natural language request for edits ("Could you check if this is positive enough?”).

[0064] As shown in Figure 21, the machine learning agent has edited the text selected by the user to updated or edited text (“I’m sure the summit will be great!”). The machine learning agent has also added a responsive comment (“I’ve changed it.”).

Example Inference Processes

[0065] Figure 3A depicts a block diagram of an example inference process for a machine-learned language model 120 according to example embodiments of the present disclosure. As shown in Figure 3A, a natural language instruction 350 and a set of textual content 352 can be provided as input to the machine-learned language model 120.

[0066] In some implementations, the set of textual content 352 can include an entirety of a document being edited in a text generation application. In other implementations, the set of textual content 352 can include one or more subportions of the document. For example, the subportions of the document may be portions that have been selected (e.g.. using highlighting) by a user. In some implementations, one or more sentences that surround the portion(s) selected by the user can also be included in the set of textual content 352 so as to provide additional context for the instructions 350.

[0067] In some implementations, the natural language instruction 350 and the set of textual content 352 can be concatenated to form a model prompt that is provided as input. In some implementations, the model prompt can be structured according to a template.

[0068] In response to the input, the machine-learned language model 120 can generate one or more edits 360 to the set of textual content 352. In some implementations, the edits 360 generated by the model 120 can be a full set of replacement text. In some implementations, the edits 360 generated by the model 120 can describe specific edits to the existing set of textual content 352. In some implementations, the edits 360 generated by the model 120 can be provided in a clean format (without edits or changes explicitly shown). In some implementations, the edits 360 generated by the model 120 can be provided in a markup format (e.g., with edits or changes shown using visual cues such as strikethrough, underlining, highlighting, and/or font color, etc.).

[0069] Figure 3B depicts a block diagram of another example inference process for a machine-learned language model 120 according to example embodiments of the present disclosure. Figure 3B is highly similar to Figure 3 A, except that in Figure 3B, the model 120 also outputs a responsive comment 362. In some implementations, the edit(s) 360 and the responsive comment 362 can be generated by the model 120 in a single inference run. In other implementations, the edit(s) 360 and the responsive comment 362 can be generated by the model 120 can be generated in different inference runs. For example, two prompts can be created: one prompt can request that the model 120 perform the requested edits 360. A second prompt can request that the model 120 generate the responsive comment 362 as a summary of the edits 360.

Example Training Processes

[0070] Figures 4A-B depict block diagrams of example training processes for a machine- learned language model 120 according to example embodiments of the present disclosure. As illustrated in Figure 4A, a computing system can obtain a training tuple 430. The training tuple 430 can include a natural language instruction 450, a set of pre-edit text 434, and a set of post-edit text 432.

[0071] The natural language instruction 450 and the pre-edit text 432 can be provided as an input to the machine-learned language model 120. In response, the machine-learned language model 120 can generate a set of predicted text 460.

[0072] The computing system can evaluate a loss function 470 that generates a loss value based on a comparison of the post-edit text 432 with the predicted text 460. The computing system can update one or more parameter values of the machine-learned language model 120 based on the loss function 470. For example, the loss function 470 can be backpropagated through the machine-learned language model 120.

[0073] The training tuple 430 can be created from a number of different sources. As one example, the training tuple 430 can be created from annotation and edit histories from the same text generation application or other productivity tool in which the machine learning model 120 is to be deployed. For example, the annotation and edit histories can include or be derived from a record or log of comments that have been added, edited, resolved, or deleted in a number of documents during a collaboration process. The histories can include a record of comments and responsive edits and/or responsive comments.

[0074] As another example, the training tuple 430 can be created from edit histories from large text corpuses, such as editing histories for an online, collaboratively-edited encyclopedia. For each edit in the edit history of the text corpus, a summary' of the edit maybe available. This summary can be included in the training tuple 430 as the natural language instruction 450. The summary of the edit can be included in the training tuple 430 as-is or can be modified or refined via application of human changes or edits made by a separate language model.

[0075] More generally, a number of pre- or post- processing techniques can also be performed on the training tuple 430 to improve the performance of the language model 120 trained on the training tuple 430. Thus, training tuples 430 can be generated which enable training the language model 120 to perform generalized text editing of long-form textual content.

[0076] Figure 4B depicts a training approach that is similar to the approach in Figure 4A. In Figure 4B, however, the training tuple 430 further includes a ground-truth response 480. In addition, the model 120 can generate a predicted response 490. The loss function 470 can further compare the ground-truth response 480 to the predicted response 490. Although a single loss function 470 is used, in other cases multiple different loss functions can be used.

Example Methods

[0077] Figure 5 depicts a flowchart of a method 500 for training one or more machine- learned models according to aspects of the present disclosure. For instance, an example machine-learned model can include a language model such as a large language model or a large multimodal model capable of processing language and one or more additional modalities.

[0078] One or more portion(s) of example method 500 can be implemented by a computing system that includes one or more computing devices such as, for example, computing systems described with reference to the other figures. Each respective portion of example method 500 can be performed by any (or any combination) of one or more computing devices. Moreover, one or more portion(s) of example method 500 can be implemented on the hardware components of the device(s) described herein, for example, to train one or more systems or models. Figure 5 depicts elements performed in a particular order for purposes of illustration and discussion. Those of ordinary skill in the art, using the disclosures provided herein, will understand that the elements of any of the methods discussed herein can be adapted, rearranged, expanded, omitted, combined, or modified in various ways without deviating from the scope of the present disclosure. Figure 5 is described with reference to elements/terms described with respect to other systems and figures for exemplary illustrated purposes and is not meant to be limiting. One or more portions of example method 500 can be performed additionally, or alternatively, by other systems. [0079] At 502, example method 500 can include obtaining a training instance. A set of training data can include a plurality of training instances divided between multiple datasets (e.g., a training dataset, a validation dataset, or testing dataset). A training instance can be labeled or unlabeled. Although referred to in example method 500 as a “training” instance, it is to be understood that runtime inferences can form training instances when a model is trained using an evaluation of the model’s performance on that runtime instance (e g., online training/leaming). Example data types for the training instance and various tasks associated therewith are described throughout the present disclosure.

[0080] At 504, example method 500 can include processing, using one or more machine- learned models, the training instance to generate an output. The output can be directly- obtained from the one or more machine-learned models or can be a downstream result of a chain of processing operations that includes an output of the one or more machine-learned models.

[0081] At 506, example method 500 can include receiving an evaluation signal associated with the output. The evaluation signal can be obtained using a loss function. Various determinations of loss can be used, such as mean squared error, likelihood loss, cross entropy loss, hinge loss, contrastive loss, or various other loss functions. The evaluation signal can be computed using known ground-truth labels (e.g., supervised learning), predicted or estimated labels (e.g., semi- or self-supervised learning), or without labels (e.g., unsupervised learning). The evaluation signal can be a reward (e.g., for reinforcement learning). The reward can be computed using a machine-learned reward model configured to generate rewards based on output(s) received. The reward can be computed using feedback data describing human feedback on the output(s). In another example, the reward model can provide a reward based on an edit ratio between the input and output and/or a Natural Language Inference score between the input and output. In another example, the reward model can be trained to predict a reward based on training data that includes training quality^ scores generated based on an edit ratio between an input and an output and/or a Natural Language Inference score between the input and output.

[0082] At 508, example method 500 can include updating the machine-learned model using the evaluation signal. For example, values for parameters of the machine-learned model(s) can be learned, in some embodiments, using various training or learning techniques, such as, for example, backwards propagation. For example, the evaluation signal can be backpropagated from the output (or another source of the evaluation signal) through the machine-learned model(s) to update one or more parameters of the model(s) (e.g., based on a gradient of the evaluation signal with respect to the parameter value(s)). For example, system(s) containing one or more machine-learned models can be trained in an end-to-end manner. Gradient descent techniques can be used to iteratively update the parameters over a number of training iterations. In some implementations, performing backwards propagation of errors can include performing truncated backpropagation through time. Example method 500 can include implementing a number of generalization techniques (e.g., weight decays, dropouts, etc.) to improve the generalization capability of the models being trained.

[0083] In some implementations, example method 500 can be implemented for training a machine-learned model from an initialized state to a fully trained state (e.g., when the model exhibits a desired performance profile, such as based on accuracy, precision, recall, etc.). [0084] In some implementations, example method 500 can be implemented for particular stages of a training procedure. For instance, in some implementations, example method 500 can be implemented for pre-training a machine-learned model. Pre-training can include, for instance, large-scale training over potentially noisy data to achieve a broad base of performance levels across a variety of tasks/data types. In some implementations, example method 500 can be implemented for fine-tuning a machine-learned model. Fine-tuning can include, for instance, smaller-scale training on higher-quality (e.g., labeled, curated, etc.) data. Fine-tuning can affect all or a portion of the parameters of a machine-learned model. For example, various portions of the machine-learned model can be ‘‘frozen’' for certain training stages. For example, parameters associated with an embedding space can be “frozen” during fine-tuning (e.g., to retain information learned from a broader domain(s) than present in the fine-tuning dataset(s)). An example fine-tuning approach includes reinforcement learning. Reinforcement learning can be based on user feedback on model performance during use.

Example Machine-Learned Models

[0085] Figure 6 is a block diagram of an example processing flow for using machine- learned model(s) 1 to process input(s) 2 to generate output(s) 3.

[0086] Machine-learned model(s) 1 can be or include one or multiple machine-learned models or model components. Example machine-learned models can include neural networks (e.g., deep neural networks). Example machine-learned models can include non-linear models or linear models. Example machine-learned models can use other architectures in lieu of or in addition to neural networks. Example machine-learned models can include decision tree based models, support vector machines, hidden Markov models, Bayesian networks, linear regression models, k-means clustering models, etc.

[0087] Example neural networks can include feed-forward neural networks, recurrent neural networks (RNNs), including long short-term memory (LSTM) based recurrent neural networks, convolutional neural networks (CNNs), diffusion models, generative-adversarial networks, or other forms of neural networks. Example neural networks can be deep neural networks. Some example machine-learned models can leverage an attention mechanism such as self-attention. For example, some example machine-learned models can include multiheaded self-attention models.

[0088] Machine-learned model(s) 1 can include a single or multiple instances of the same model configured to operate on data from input(s) 2. Machine-learned model(s) 1 can include an ensemble of different models that can cooperatively interact to process data from input(s) 2. For example, machine-learned model(s) 1 can employ a mixture-of-experts structure. See, e.g., Zhou et al., Mixture-of-Experts with Expert Choice Routing, ARXIV:2202.09368V2 (Oct. 14, 2022).

[0089] Input(s) 2 can generally include or otherwise represent various types of data. Input(s) 2 can include one type or many different types of data. Output(s) 3 can be data of the same type(s) or of different types of data as compared to input(s) 2. Output(s) 3 can include one type or many different types of data.

[0090] Example data types for input(s) 2 or output(s) 3 include natural language text data, software code data (e.g., source code, object code, machine code, or any other form of computer-readable instructions or programming languages), machine code data (e.g., binary code, assembly code, or other forms of machine-readable instructions that can be executed directly by a computer's central processing unit), assembly code data (e.g., low-level programming languages that use symbolic representations of machine code instructions to program a processing unit), genetic data or other chemical or biochemical data, image data, audio data, audiovisual data, haptic data, biometric data, medical data, financial data, statistical data, geographical data, astronomical data, historical data, sensor data generally (e.g.. digital or analog values, such as voltage or other absolute or relative level measurement values from a real or artificial input, such as from an audio sensor, light sensor, displacement sensor, etc.), and the like. Data can be raw or processed and can be in any format or schema. [0091] In multimodal inputs 2 or outputs 3, example combinations of data types include image data and audio data, image data and natural language data, natural language data and software code data, image data and biometric data, sensor data and medical data, etc. It is to be understood that any combination of data types in an input 2 or an output 3 can be present. [0092] An example input 2 can include one or multiple data types, such as the example data types noted above. An example output 3 can include one or multiple data types, such as the example data types noted above. The data type(s) of input 2 can be the same as or different from the data type(s) of output 3. It is to be understood that the example data types noted above are provided for illustrative purposes only. Data types contemplated within the scope of the present disclosure are not limited to those examples noted above.

Example Machine-Learned Sequence Processing Models

[0093] Figure 7 is a block diagram of an example implementation of an example machine-learned model configured to process sequences of information. For instance, an example implementation of machine-learned model(s) 1 can include machine-learned sequence processing model(s) 4. An example system can pass input(s) 2 to sequence processing model(s) 4. Sequence processing model(s) 4 can include one or more machine- learned components. Sequence processing model(s) 4 can process the data from input(s) 2 to obtain an input sequence 5. Input sequence 5 can include one or more input elements 5-1, 5- 2, . . . , 5-AL, etc. obtained from input(s) 2. Sequence processing model 4 can process input sequence 5 using prediction layer(s) 6 to generate an output sequence 7. Output sequence 7 can include one or more output elements 7-1, 7-2. . . . , 7-N. etc. generated based on input sequence 5. The system can generate output(s) 3 based on output sequence 7.

[0094] Sequence processing model(s) 4 can include one or multiple machine-learned model components configured to ingest, generate, or otherwise reason over sequences of information. For example, some example sequence processing models in the text domain are referred to as “Large Language Models,’’ or LLMs. See, e.g., PaLM 2 Technical Report, GOOGLE, https://ai.google/static/documents/palm2techreport.pdf (n.d.). Other example sequence processing models can operate in other domains, such as image domains, see. e.g., Dosovitskiy et al., An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale, ARXIV:2010.11929V2 (Jun. 3, 2021), audio domains, see. e.g., Agostinelli et al., MusicLM: Generating Music From Text. ARXlV:2301.11325vl (Jan. 26, 2023), biochemical domains, see, e.g., Jumper et al., Highly accurate protein structure prediction with AlphaFold, 596 Nature 583 (Aug. 26, 2021), by way of example. Sequence processing model(s) 4 can process one or multiple types of data simultaneously. Sequence processing model(s) 4 can include relatively large models (e.g.. more parameters, computationally expensive, etc.), relatively small models (e.g., fewer parameters, computationally lightweight, etc.), or both. [0095] In general, sequence processing model(s) 4 can obtain input sequence 5 using data from input(s) 2. For instance, input sequence 5 can include a representation of data from input(s) 2 in a format understood by sequence processing model(s) 4. One or more machine- learned components of sequence processing model(s) 4 can ingest the data from input(s) 2, parse the data into pieces compatible with the processing architectures of sequence processing model(s) 4 (e.g., via “tokenization”), and project the pieces into an input space associated with prediction layer(s) 6 (e.g., via “embedding”).

[0096] Sequence processing model(s) 4 can ingest the data from input(s) 2 and parse the data into a sequence of elements to obtain input sequence 5. For example, a portion of input data from input(s) 2 can be broken down into pieces that collectively represent the content of the portion of the input data. The pieces can provide the elements of the sequence.

[0097] Elements 5-1, 5-2, . . . , 5-M can represent, in some cases, building blocks for capturing or expressing meaningful information in a particular data domain. For instance, the elements can describe “atomic units” across one or more domains. For example, for textual input source(s), the elements can correspond to groups of one or more words or sub-word components, such as sets of one or more characters.

[0098] For example, elements 5-1, 5-2, . . . , 5-M can represent tokens obtained using a tokenizer. For instance, a tokenizer can process a given portion of an input source and output a series of tokens (e g., corresponding to input elements 5-1 , 5-2, . . . , 5-M) that represent the portion of the input source. Various approaches to tokenization can be used. For instance, textual input source(s) can be tokenized using a byte-pair encoding (BPE) technique. See, e.g., Kudo et al., SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing, PROCEEDINGS OF THE 2018 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING (System Demonstrations), pages 66-71 (October 31-November 4, 2018), https://aclanthology .org/D18-2012.pdf. Image-based input source(s) can be tokenized by extracting and serializing patches from an image.

[0099] In general, arbitrary data types can be serialized and processed into input sequence 5. It is to be understood that element(s) 5-1, 5-2, . . . , 5-M depicted in Figure 7 can be the tokens or can be the embedded representations thereof.

[0100] Prediction layer(s) 6 can predict one or more output elements 7-1, 7-2, . . . , 1-N based on the input elements. Prediction layer(s) 6 can include one or more machine-learned model architectures, such as one or more layers of learned parameters that manipulate and transform the input(s) to extract higher-order meaning from, and relationships between, input element(s) 5-1. 5-2, . . . . 5-M. In this manner, for instance, example prediction layer(s) 6 can predict new output element(s) in view of the context provided by input sequence 5.

[0101] Prediction layer(s) 6 can evaluate associations between portions of input sequence 5 and a particular output element. These associations can inform a prediction of the likelihood that a particular output follows the input context. For example, consider the textual snippet, “The carpenter’s toolbox was small and heavy. It was full of Example prediction layer(s) 6 can identify that “It” refers back to “toolbox” by determining a relationship between the respective embeddings. Example prediction layer(s) 6 can also link “It” to the attributes of the toolbox, such as “small” and “heavy.” Based on these associations, prediction layer(s) 6 can, for instance, assign a higher probability to the word “nails” than to the word “sawdust.”

[0102] A transformer is an example architecture that can be used in prediction layer(s) 4. See, e.g., Vaswani et al., Attention Is All You Need, ARXIV: 1706.03762V7 (Aug. 2, 2023). A transformer is an example of a machine-learned model architecture that uses an attention mechanism to compute associations between items within a context window. The context window can include a sequence that contains input sequence 5 and potentially one or more output element(s) 7-1, 7-2, . . . , 7-N. A transformer block can include one or more attention layer(s) and one or more post-attention layer(s) (e g., feedforward layer(s), such as a multilayer perceptron).

[0103] Prediction layer(s) 6 can include other machine-learned model architectures in addition to or in lieu of transformer-based architectures. For example, recurrent neural networks (RNNs) and long short-term memory' (LSTM) models can also be used, as well as convolutional neural networks (CNNs). In general, prediction layer(s) 6 can leverage various kinds of artificial neural networks that can understand or generate sequences of information. [0104] Output sequence 7 can include or otherwise represent the same or different data ty pes as input sequence 5. For instance, input sequence 5 can represent textual data, and output sequence 7 can represent textual data. Input sequence 5 can represent image, audio, or audiovisual data, and output sequence 7 can represent textual data (e.g., describing the image, audio, or audiovisual data). It is to be understood that prediction layer(s) 6, and any other interstitial model components of sequence processing model(s) 4, can be configured to receive a variety of data types in input sequence(s) 5 and output a variety of data ty pes in output sequence(s) 7. [0105] Output sequence 7 can have various relationships to input sequence 5. Output sequence 7 can be a continuation of input sequence 5. Output sequence 7 can be complementary to input sequence 5. Output sequence 7 can translate, transform, augment, or otherw ise modify input sequence 5. Output sequence 7 can answer, evaluate, confirm, or otherwise respond to input sequence 5. Output sequence 7 can implement (or describe instructions for implementing) an instruction provided via input sequence 5.

[0106] Output sequence 7 can be generated autoregressively. For instance, for some applications, an output of one or more prediction layer(s) 6 can be passed through one or more output layers (e.g., softmax layer) to obtain a probability distribution over an output vocabulary (e.g., a textual or symbolic vocabulary) conditioned on a set of input elements in a context window. In this manner, for instance, output sequence 7 can be autoregressively generated by sampling a likely next output element, adding that element to the context window, and re-generating the probability distribution based on the updated context window and sampling a likely next output element, and so forth.

[0107] Output sequence 7 can also be generated non-autoregressively. For instance, multiple output elements of output sequence 7 can be predicted together without explicit sequential conditioning on each other. See, e.g., Saharia et al., Non-Autoregressive Machine Translation with Latent Alignments, ARXlV:2004.07437v3 (NOV. 16, 2020).

[0108] Output sequence 7 can include one or multiple portions or elements. In an example content generation configuration, output sequence 7 can include multiple elements corresponding to multiple portions of a generated output sequence (e.g., a textual sentence, values of a discretized w aveform, computer code, etc.). In an example classification configuration, output sequence 7 can include a single element associated with a classification output. For instance, an output “vocabulary” can include a set of classes into which an input sequence is to be classified. For instance, a vision transformer block can pass latent state information to a multilayer perceptron that outputs a likely class value associated with an input image.

[0109] Figure 8 is a block diagram of an example technique for populating an example input sequence 8. Input sequence 8 can include various functional elements that form part of the model infrastructure, such as an element 8-0 obtained from a task indicator 9 that signals to any model(s) that process input sequence 8 that a particular task is being performed (e.g., to help adapt a performance of the model(s) to that particular task). Input sequence 8 can include various data elements from different data modalities. For instance, an input modality 10-1 can include one modality of data. A data-to-sequence model 11-1 can process data from input modality 10-1 to project the data into a format compatible with input sequence 8 (e.g., one or more vectors dimensioned according to the dimensions of input sequence 8) to obtain elements 8-1, 8-2, 8-3. Another input modality 10-2 can include a different modality of data. A data-to-sequence model 11-2 can project data from input modality 10-2 into a format compatible with input sequence 8 to obtain elements 8-4, 8-5, 8-6. Another input modality 10-3 can include yet another different modality of data. A data-to-sequence model 11-3 can project data from input modality 10-3 into a format compatible with input sequence 8 to obtain elements 8-7, 8-8, 8-9.

[0110] Input sequence 8 can be the same as or different from input sequence 5. Input sequence 8 can be a multimodal input sequence that contains elements that represent data from different modalities using a common dimensional representation. For instance, an embedding space can have P dimensions. Input sequence 8 can be configured to contain a plurality of elements that have P dimensions. In this manner, for instance, example implementations can facilitate information extraction and reasoning across diverse data modalities by projecting data into elements in the same embedding space for comparison, combination, or other computations therebetween.

[0111] For example, elements 8-0, . . . , 8-9 can indicate particular locations within a multidimensional embedding space. Some elements can map to a set of discrete locations in the embedding space. For instance, elements that correspond to discrete members of a predetermined vocabulary of tokens can map to discrete locations in the embedding space that are associated with those tokens. Other elements can be continuously distributed across the embedding space. For instance, some datatypes can be broken down into continuously defined portions (e.g., image patches) that can be described using continuously distributed locations within the embedding space.

[0112] In some implementations, the expressive power of the embedding space may not be limited to meanings associated with any particular set of tokens or other building blocks. For example, a continuous embedding space can encode a spectrum of high-order information. An individual piece of information (e.g., a token) can map to a particular point in that space: for instance, a token for the word ‘"dog” can be projected to an embedded value that points to a particular location in the embedding space associated with canine-related information. Similarly, an image patch of an image of a dog on grass can also be projected into the embedding space. In some implementations, the projection of the image of the dog can be similar to the projection of the word “dog” while also having similarity to a projection of the word “grass,” while potentially being different from both. In some implementations. the projection of the image patch may not exactly align with any single projection of a single word. In some implementations, the projection of the image patch can align with a combination of the projections of the words '‘dog’’ and ‘'grass.” In this manner, for instance, a high-order embedding space can encode information that can be independent of data modalities in which the information is expressed.

[0113] Task indicator 9 can include a model or model component configured to identify a task being performed and inject, into input sequence 8, an input value represented by element 8-0 that signals which task is being performed. For instance, the input value can be provided as a data type associated with an input modality and projected along with that input modality (e.g., the input value can be a textual task label that is embedded along with other textual data in the input; the input value can be a pixel-based representation of a task that is embedded along with other image data in the input; etc.). The input value can be provided as a data type that differs from or is at least independent from other input(s). For instance, the input value represented by element 8-0 can be a learned within a continuous embedding space.

[0114] Input modalities 10-1. 10-2. and 10-3 can be associated with various different data types (e.g., as described above with respect to input(s) 2 and output(s) 3).

[0115] Data-to-sequence models 11-1, 11-2, and 11-3 can be the same or different from each other. Data-to-sequence models 11-1, 11-2, and 11-3 can be adapted to each respective input modality 10-1, 10-2, and 10-3. For example, a textual data-to-sequence model can subdivide a portion of input text and project the subdivisions into element(s) in input sequence 8 (e.g., elements 8-1, 8-2, 8-3, etc.). An image data-to-sequence model can subdivide an input image and project the subdivisions into element(s) in input sequence 8 (e.g., elements 8-4. 8-5, 8-6, etc.). An arbitrary datatype data-to-sequence model can subdivide an input of that arbitrary datatype and project the subdivisions into element(s) in input sequence 8 (e.g., elements 8-7, 8-8, 8-9, etc.).

[0116] Data-to-sequence models 11-1, 11-2, and 11-3 can form part of machine-learned sequence processing model(s) 4. Data-to-sequence models 11-1, 11-2, and 11-3 can be jointly trained with or trained independently from machine-learned sequence processing model(s) 4. Data-to-sequence models 11-1, 11-2, and 11-3 can be trained end-to-end with machine- learned sequence processing model(s) 4. Example Machine-Learned Model Development Platform

[0117] Figure 9 is a block diagram of an example model development platform 12 that can facilitate creation, adaptation, and refinement of example machine-learned models (e.g., machine-learned model(s) 1, sequence processing model(s) 4, etc.). Model development platform 12 can provide a number of different toolkits that developer systems can employ in the development of new or adapted machine-learned models.

[0118] Model development platform 12 can provide one or more model libraries 13 containing building blocks for new models. Model libraries 13 can include one or more pretrained foundational models 13-1, which can provide a backbone of processing power across various tasks. Model libraries 13 can include one or more pre-trained expert models 13-2, which can be focused on performance in particular domains of expertise. Model libraries 13 can include various model primitives 13-3, which can provide low-level architectures or components (optionally pre-trained), which can be assembled in various arrangements as desired.

[0119] Model development platform 12 can receive selections of various model components 14. Model development platform 12 can pass selected model components 14 to a workbench 15 that combines selected model components 14 into a development model 16. [0120] Workbench 15 can facilitate further refinement and adaptation of development model 16 by leveraging a number of different toolkits integrated with model development platform 12. For example, workbench 15 can facilitate alignment of the development model 16 with a desired performance profile on various tasks using a model alignment toolkit 17. [0121] Model alignment toolkit 17 can provide anumber of tools for causing development model 16 to generate outputs aligned with desired behavioral characteristics. Alignment can include increasing an accuracy, precision, recall, etc. of model outputs.

Alignment can include enforcing output styles, schema, or other preferential characteristics of model outputs. Alignment can be general or domain-specific. For instance, a pre-trained foundational model 13-1 can begin with an initial level of performance across multiple domains. Alignment of the pre-trained foundational model 13-1 can include improving a performance in a particular domain of information or tasks (e.g., even at the expense of performance in another domain of information or tasks).

[0122] Model alignment toolkit 17 can integrate one or more dataset(s) 17-1 for aligning development model 16. Curated dataset(s) 17-1 can include labeled or unlabeled training data. Dataset(s) 17-1 can be obtained from public domain datasets. Dataset(s) 17-1 can be obtained from private datasets associated with one or more developer system(s) for the alignment of bespoke machine-learned model(s) customized for private use-cases.

[0123] Pre-training pipelines 17-2 can include a machine-learned model training workflow configured to update development model 16 over large-scale, potentially noisy datasets. For example, pre-training can leverage unsupervised learning techniques (e.g., denoising, etc.) to process large numbers of training instances to update model parameters from an initialized state and achieve a desired baseline performance. Pre-training pipelines 17-2 can leverage unlabeled datasets in dataset(s) 17-1 to perform pre-training. Workbench 15 can implement a pre-training pipeline 17-2 to pre-train development model 16.

[0124] Fine-tuning pipelines 17-3 can include a machine-learned model training workflow configured to refine the model parameters of development model 16 with higher- quality data. Fine-tuning pipelines 17-3 can update development model 16 by conducting supervised training with labeled dataset(s) in dataset(s) 17-1. Fine-tuning pipelines 17-3 can update development model 16 by conducting reinforcement learning using reward signals from user feedback signals. Workbench 15 can implement a fine-tuning pipeline 17-3 to finetune development model 16.

[0125] Prompt libraries 17-4 can include sets of inputs configured to induce behavior aligned with desired performance criteria. Prompt libraries 17-4 can include few-shot prompts (e.g., inputs providing examples of desired model outputs for prepending to a desired runtime query), chain-of-thought prompts (e.g., inputs providing step-by-step reasoning within the exemplars to facilitate thorough reasoning by the model), and the like.

[0126] Example prompts can be retrieved from an available repository' of prompt libraries 17-4. Example prompts can be contributed by one or more developer systems using workbench 15.

[0127] In some implementations, pre-trained or fine-tuned models can achieve satisfactory' performance without exemplars in the inputs. For instance, zero-shot prompts can include inputs that lack exemplars. Zero-shot prompts can be within a domain within a training dataset or outside of the training domain(s).

[0128] Prompt libraries 17-4 can include one or more prompt engineering tools. Prompt engineering tools can provide workflows for retrieving or learning optimized prompt values. Prompt engineering tools can facilitate directly learning prompt values (e.g., input element values) based one or more training iterations. Workbench 15 can implement prompt engineering tools in development model 16. [0129] Prompt libraries 17-4 can include pipelines for prompt generation. For example, inputs can be generated using development model 16 itself or other machine-learned models. In this manner, for instance, a first model can process information about a task and output a input for a second model to process in order to perform a step of the task. The second model can be the same as or different from the first model. Workbench 15 can implement prompt generation pipelines in development model 16.

[0130] Prompt libraries 17-4 can include pipelines for context injection. For instance, a performance of development model 16 on a particular task can improve if provided with additional context for performing the task. Prompt libraries 17-4 can include software components configured to identify desired context, retrieve the context from an external source (e.g., a database, a sensor, etc.), and add the context to the input prompt. Workbench 15 can implement context injection pipelines in development model 16.

[0131] Although various training examples described herein with respect to model development platform 12 refer to “pre-training’" and “fine-tuning,” it is to be understood that model alignment toolkit 17 can generally support a wide variety of training techniques adapted for training a wide variety of machine-learned models. Example training techniques can correspond to the example training method 500 described above.

[0132] Model development platform 12 can include a model plugin toolkit 18. Model plugin toolkit 18 can include a variety of tools configured for augmenting the functionality of a machine-learned model by integrating the machine-learned model with other systems, devices, and software components. For instance, a machine-learned model can use tools to increase performance qualify where appropriate. For instance, deterministic tasks can be offloaded to dedicated tools in lieu of probabilistically performing the task with an increased risk of error. For instance, instead of autoregressively predicting the solution to a system of equations, a machine-learned model can recognize a tool to call for obtaining the solution and pass the system of equations to the appropriate tool. The tool can be a traditional system of equations solver that can operate deterministically to resolve the system of equations. The output of the tool can be returned in response to the original query. In this manner, tool use can allow some example models to focus on the strengths of machine-learned models — e.g.. understanding an intent in an unstructured request for a task — while augmenting the performance of the model by offloading certain tasks to a more focused tool for rote application of deterministic algorithms to a well-defined problem.

[0133] Model plugin toolkit 18 can include validation tools 18-1. Validation tools 18-1 can include tools that can parse and confirm output(s) of a machine-learned model. Validation tools 18-1 can include engineered heuristics that establish certain thresholds applied to model outputs. For example, validation tools 18-1 can ground the outputs of machine-learned models to structured data sources (e.g., to mitigate “hallucinations”). [0134] Model plugin toolkit 18 can include tooling packages 18-2 for implementing one or more tools that can include scripts or other executable code that can be executed alongside development model 16. Tooling packages 18-2 can include one or more inputs configured to cause machine-learned model(s) to implement the tools (e.g., few-shot prompts that induce a model to output tool calls in the proper syntax, etc.). Tooling packages 18-2 can include, for instance, fine-tuning training data for training a model to use a tool.

[0135] Model plugin toolkit 18 can include interfaces for calling external application programming interfaces (APIs) 18-3. For instance, in addition to or in lieu of implementing tool calls or tool code directly with development model 16, development model 16 can be aligned to output instruction that initiate API calls to send or obtain data via external systems. [0136] Model plugin toolkit 18 can integrate with prompt libraries 17-4 to build a catalog of available tools for use with development model 16. For instance, a model can receive, in an input, a catalog of available tools, and the model can generate an output that selects a tool from the available tools and initiates a tool call for using the tool.

[0137] Model development platform 12 can include a computational optimization toolkit 19 for optimizing a computational performance of development model 16. For instance, tools for model compression 19-1 can allow development model 16 to be reduced in size while maintaining a desired level of performance. For instance, model compression 19-1 can include quantization workflows, weight pruning and sparsification techniques, etc. Tools for hardware acceleration 19-2 can facilitate the configuration of the model storage and execution formats to operate optimally on different hardware resources. For instance, hardware acceleration 19-2 can include tools for optimally sharding models for distributed processing over multiple processing units for increased bandwidth, lower unified memory requirements, etc. Tools for distillation 19-3 can provide for the training of lighter-weight models based on the knowledge encoded in development model 16. For instance, development model 16 can be a highly performant, large machine-learned model optimized using model development platform 12. To obtain a lightweight model for running in resource-constrained environments, a smaller model can be a “student model” that learns to imitate development model 16 as a “teacher model.” In this manner, for instance, the investment in learning the parameters and configurations of development model 16 can be efficiently transferred to a smaller model for more efficient inference. [0138] Workbench 15 can implement one, multiple, or none of the toolkits implemented in model development platform 12. Workbench 15 can output an output model 20 based on development model 16. Output model 20 can be a deployment version of development model 16. Output model 20 can be a development or training checkpoint of development model 16. Output model 20 can be a distilled, compressed, or otherw ise optimized version of development model 16.

[0139] Figure 10 is a block diagram of an example training flow for training a machine- learned development model 16. One or more portion(s) of the example training flow can be implemented by a computing system that includes one or more computing devices such as, for example, computing systems described with reference to the other figures. Each respective portion of the example training flow can be performed by any (or any combination) of one or more computing devices. Moreover, one or more portion(s) of the example training flow can be implemented on the hardware components of the device(s) described herein, for example, to train one or more systems or models. FIG. 10 depicts elements performed in a particular order for purposes of illustration and discussion. Those of ordinary skill in the art, using the disclosures provided herein, will understand that the elements of any of the methods discussed herein can be adapted, rearranged, expanded, omitted, combined, or modified in various w ays w ithout deviating from the scope of the present disclosure. FIG. 10 is described with reference to elements/terms described with respect to other systems and figures for exemplary illustrated purposes and is not meant to be limiting. One or more portions of the example training flow' can be performed additionally, or alternatively, by other systems.

[0140] Initially, development model 16 can persist in an initial state as an initialized model 21. Development model 16 can be initialized with weight values. Initial weight values can be random or based on an initialization schema. Initial weight values can be based on prior pre-training for the same or for a different model.

[0141] Initialized model 21 can undergo pre-training in a pre-training stage 22. Pretraining stage 22 can be implemented using one or more pre-training pipelines 17-2 over data from dataset(s) 17-1. Pre-training can be omitted, for example, if initialized model 21 is already pre-trained (e g., development model 16 contains, is, or is based on a pre-trained foundational model or an expert model).

[0142] Pre-trained model 23 can then be a new version of development model 16, which can persist as development model 16 or as anew development model. Pre-trained model 23 can be the initial state if development model 16 w as already pre-trained. Pre-trained model 23 can undergo fine-tuning in a fine-tuning stage 24. Fine-tuning stage 24 can be implemented using one or more fine-tuning pipelines 17-3 over data from dataset(s) 17-1. Fine-tuning can be omitted, for example, if a pre-trained model as satis factory performance, if the model was already fine-tuned, or if other tuning approaches are preferred.

[0143] Fine-tuned model 29 can then be a new version of development model 16, which can persist as development model 16 or as anew development model. Fine-tuned model 29 can be the initial state if development model 16 was already fine-tuned. Fine-tuned model 29 can undergo refinement with user feedback 26. For instance, refinement with user feedback 26 can include reinforcement learning, optionally based on human feedback from human users of fine-tuned model 25. As reinforcement learning can be a form of fine-tuning, it is to be understood that fine-tuning stage 24 can subsume the stage for refining with user feedback 26. Refinement with user feedback 26 can produce a refined model 27. Refined model 27 can be output to downstream system(s) 28 for deployment or further development.

[0144] In some implementations, computational optimization operations can be applied before, during, or after each stage. For instance, initialized model 21 can undergo computational optimization 29-1 (e.g.. using computational optimization toolkit 19) before pre-training stage 22. Pre-trained model 23 can undergo computational optimization 29-2 (e.g., using computational optimization toolkit 19) before fine-tuning stage 24. Fine-tuned model 25 can undergo computational optimization 29-3 (e.g., using computational optimization toolkit 19) before refinement with user feedback 26. Refined model 27 can undergo computational optimization 29-4 (e.g., using computational optimization toolkit 19) before output to downstream system(s) 28. Computational optimization(s) 29-1, . . . , 29-4 can all be the same, all be different, or include at least some different optimization techniques.

Example Machine-Learned Model Inference System

[0145] Figure 11 is a block diagram of an inference system for operating one or more machine-learned model(s) 1 to perform inference (e.g., for training, for deployment, etc.). A model host 31 can receive machine-learned model(s) 1. Model host 31 can host one or more model instance(s) 31-1, which can be one or multiple instances of one or multiple models. Model host 31 can host model instance(s) 31-1 using available compute resources 31-2 associated with model host 31.

[0146] Model host 31 can perform inference on behalf of one or more client(s) 32. Client(s) 32 can transmit an input request 33 to model host 31. Using input request 33, model host 31 can obtain input(s) 2 for input to machine-learned model(s) 1. Machine-learned model(s) 1 can process input(s) 2 to generate output(s) 3. Using output(s) 3, model host 31 can return an output payload 34 for responding to input request 33 from client(s) 32. Output payload 34 can include or be based on output(s) 3.

[0147] Model host 31 can leverage various other resources and tools to augment the inference task. For instance, model host 31 can communicate with tool interfaces 35 to facilitate tool use by model instance(s) 31-1. Tool interfaces 35 can include local or remote APIs. Tool interfaces 35 can include integrated scripts or other software functionality. Model host 31 can engage online learning interface(s) 36 to facilitate ongoing improvements to machine-learned model(s) 1. For instance, online learning interface(s) 36 can be used within reinforcement learning loops to retrieve user feedback on inferences served by model host 31. Model host 31 can access runtime data source(s) 37 for augmenting input(s) 2 with additional contextual information. For instance, runtime data source(s) 37 can include a knowledge graph 37-1 that facilitates structured information retrieval for information associated with input request(s) 33 (e.g., a search engine service). Runtime data source(s) 37 can include public or private, external or local database(s) 37-2 that can store information associated with input request(s) 33 for augmenting input(s) 2. Runtime data source(s) 37 can include account data 37-3 which can be retrieved in association with a user account corresponding to a client 32 for customizing the behavior of model host 31 accordingly.

[0148] Model host 31 can be implemented by one or multiple computing devices or systems. Client(s) 2 can be implemented by one or multiple computing devices or systems, which can include computing devices or systems shared with model host 31.

[0149] For example, model host 31 can operate on a server system that provides a machine-learning service to client device(s) that operate client(s) 32 (e.g., over a local or wide-area network). Client device(s) can be end-user devices used by individuals. Client device(s) can be server systems that operate client(s) 32 to provide various functionality as a senice to downstream end-user devices.

[0150] In some implementations, model host 31 can operate on a same device or system as client(s) 32. Model host 31 can be a machine-learning service that runs on-device to provide machine-learning functionality to one or multiple applications operating on a client device, which can include an application implementing client(s) 32. Model host 31 can be a part of a same application as client(s) 32. For instance, model host 31 can be a subroutine or method implemented by one part of an application, and client(s) 32 can be another subroutine or method that engages model host 31 to perform inference functions within the application. It is to be understood that model host 31 and client(s) 32 can have various different configurations.

[0151] Model instance(s) 31-1 can include one or more machine-learned models that are available for performing inference. Model instance(s) 31-1 can include weights or other model components that are stored on in persistent storage, temporarily cached, or loaded into high-speed memory. Model instance(s) 31-1 can include multiple instance(s) of the same model (e.g., for parallel execution of more requests on the same model). Model instance(s) 31-1 can include instance(s) of different model(s). Model instance(s) 31-1 can include cached intermediate states of active or inactive model(s) used to accelerate inference of those models. For instance, an inference session with a particular model may generate significant amounts of computational results that can be re-used for future inference runs (e.g., using a KV cache for transformer-based models). These computational results can be saved in association with that inference session so that session can be executed more efficiently when resumed.

[0152] Compute resource(s) 31-2 can include one or more processors (central processing units, graphical processing units, tensor processing units, machine-learning accelerators, etc.) connected to one or more memory devices. Compute resource(s) 31-2 can include a dynamic pool of available resources shared with other processes. Compute resource(s) 31-2 can include memory devices large enough to fit an entire model instance in a single memory instance. Compute resource(s) 31-2 can also shard model instance(s) across multiple memory devices (e g., using data parallelization or tensor parallelization, etc.). This can be done to increase parallelization or to execute a large model using multiple memory devices which individually might not be able to fit the entire model into memory.

[0153] Input request 33 can include data for input(s) 2. Model host 31 can process input request 33 to obtain input(s) 2. Input(s) 2 can be obtained directly from input request 33 or can be retrieved using input request 33. Input request 33 can be submitted to model host 31 via an API.

[0154] Model host 31 can perform inference over batches of input requests 33 in parallel. For instance, a model instance 31-1 can be configured with an input structure that has a batch dimension. Separate input(s) 2 can be distributed across the batch dimension (e.g., rows of an array). The separate input(s) 2 can include completely different contexts. The separate input(s) 2 can be multiple inference steps of the same task. The separate input(s) 2 can be staggered in an input structure, such that any given inference cycle can be operating on different portions of the respective input(s) 2. In this manner, for instance, model host 31 can perform inference on the batch in parallel, such that output(s) 3 can also contain the batch dimension and return the inference results for the batched input(s) 2 in parallel. In this manner, for instance, batches of input request(s) 33 can be processed in parallel for higher throughput of output payload(s) 34.

[0155] Output pay load 34 can include or be based on output(s) 3 from machine-learned model(s) 1. Model host 31 can process output(s) 3 to obtain output payload 34. This can include chaining multiple rounds of inference (e.g., iteratively, recursively, across the same model(s) or different model(s)) to arrive at a final output for a task to be returned in output payload 34. Output payload 34 can be transmitted to client(s) 32 via an API.

[0156] Online learning interface(s) 36 can facilitate reinforcement learning of machine- learned model(s) 1. Online learning interface(s) 36 can facilitate reinforcement learning with human feedback (RLHF). Online learning interface(s) 36 can facilitate federated learning of machine-learned model(s) 1.

[0157] Model host 31 can execute machine-learned model(s) 1 to perform inference for various tasks using various types of data. For example, various different input(s) 2 and output(s) 3 can be used for various different tasks. In some implementations, input(s) 2 can be or otherwise represent image data. Machine-learned model(s) 1 can process the image data to generate an output. As an example, machine-learned model(s) 1 can process the image data to generate an image recognition output (e.g., a recognition of the image data, a latent embedding of the image data, an encoded representation of the image data, a hash of the image data, etc.). As another example, machine-learned model(s) 1 can process the image data to generate an image segmentation output. As another example, machine-learned model(s) 1 can process the image data to generate an image classification output. As another example, machine-learned model(s) 1 can process the image data to generate an image data modification output (e.g., an alteration of the image data, etc.). As another example, machine- learned model(s) 1 can process the image data to generate an encoded image data output (e.g., an encoded and/or compressed representation of the image data, etc.). As another example, machine-learned model(s) 1 can process the image data to generate an upscaled image data output. As another example, machine-learned model(s) 1 can process the image data to generate a prediction output.

[0158] In some implementations, the task is a computer vision task. In some cases, input(s) 2 includes pixel data for one or more images and the task is an image processing task. For example, the image processing task can be image classification, where the output is a set of scores, each score corresponding to a different object class and representing the likelihood that the one or more images depict an object belonging to the object class. The image processing task may be object detection, where the image processing output identifies one or more regions in the one or more images and, for each region, a likelihood that region depicts an object of interest. As another example, the image processing task can be image segmentation, where the image processing output defines, for each pixel in the one or more images, a respective likelihood for each category in a predetermined set of categories. For example, the set of categories can be foreground and background. As another example, the set of categories can be object classes. As another example, the image processing task can be depth estimation, where the image processing output defines, for each pixel in the one or more images, a respective depth value. As another example, the image processing task can be motion estimation, where the network input includes multiple images, and the image processing output defines, for each pixel of one of the input images, a motion of the scene depicted at the pixel between the images in the network input.

[0159] In some implementations, input(s) 2 can be or otherwise represent natural language data. Machine-learned model(s) 1 can process the natural language data to generate an output. As an example, machine-learned model(s) 1 can process the natural language data to generate a language encoding output. As another example, machine-learned model(s) 1 can process the natural language data to generate a latent text embedding output. As another example, machine-learned model(s) 1 can process the natural language data to generate a translation output. As another example, machine-learned model(s) 1 can process the natural language data to generate a classification output. As another example, machine-learned model(s) 1 can process the natural language data to generate a textual segmentation output. As another example, machine-learned model(s) 1 can process the natural language data to generate a semantic intent output. As another example, machine-learned model(s) 1 can process the natural language data to generate an upscaled text or natural language output (e.g., text or natural language data that is higher quality than the input text or natural language, etc.). As another example, machine-learned model(s) 1 can process the natural language data to generate a prediction output (e.g., one or more predicted next portions of natural language content).

[0160] In some implementations, input(s) 2 can be or otherwise represent speech data (e.g., data describing spoken natural language, such as audio data, textual data, etc.). Machine-learned model(s) 1 can process the speech data to generate an output. As an example, machine-learned model(s) 1 can process the speech data to generate a speech recognition output. As another example, machine-learned model(s) 1 can process the speech data to generate a speech translation output. As another example, machine-learned model(s) 1 can process the speech data to generate a latent embedding output. As another example, machine-learned model(s) 1 can process the speech data to generate an encoded speech output (e.g., an encoded and/or compressed representation of the speech data, etc.). As another example, machine-learned model(s) 1 can process the speech data to generate an upscaled speech output (e.g., speech data that is higher quality than the input speech data, etc.). As another example, machine-learned model(s) 1 can process the speech data to generate a textual representation output (e.g., a textual representation of the input speech data, etc.). As another example, machine-learned model(s) 1 can process the speech data to generate a prediction output.

[0161] In some implementations, input(s) 2 can be or otherwise represent latent encoding data (e.g., a latent space representation of an input, etc.). Machine-learned model(s) 1 can process the latent encoding data to generate an output. As an example, machine- learned model(s) 1 can process the latent encoding data to generate a recognition output. As another example, machine-learned model(s) 1 can process the latent encoding data to generate a reconstruction output. As another example, machine-learned model(s) 1 can process the latent encoding data to generate a search output. As another example, machine- learned model(s) 1 can process the latent encoding data to generate a reclustering output. As another example, machine-learned model(s) 1 can process the latent encoding data to generate a prediction output.

[01 2] In some implementations, input(s) 2 can be or otherwise represent statistical data. Statistical data can be, represent, or otherwise include data computed and/or calculated from some other data source. Machine-learned model(s) 1 can process the statistical data to generate an output. As an example, machine-learned model(s) 1 can process the statistical data to generate a recognition output. As another example, machine-learned model(s) 1 can process the statistical data to generate a prediction output. As another example, machine- learned model(s) 1 can process the statistical data to generate a classification output. As another example, machine-learned model(s) 1 can process the statistical data to generate a segmentation output. As another example, machine-learned model(s) 1 can process the statistical data to generate a visualization output. As another example, machine-learned model(s) 1 can process the statistical data to generate a diagnostic output.

[0163] In some implementations, input(s) 2 can be or otherwise represent sensor data. Machine-learned model(s) 1 can process the sensor data to generate an output. As an example, machine-learned model(s) 1 can process the sensor data to generate a recognition output. As another example, machine-learned model(s) 1 can process the sensor data to generate a prediction output. As another example, machine-learned model(s) 1 can process the sensor data to generate a classification output. As another example, machine-learned model(s) 1 can process the sensor data to generate a segmentation output. As another example, machine-learned model(s) 1 can process the sensor data to generate a visualization output. As another example, machine-learned model(s) 1 can process the sensor data to generate a diagnostic output. As another example, machine-learned model(s) 1 can process the sensor data to generate a detection output.

[0164] In some implementations, machine-learned model(s) 1 can be configured to perform a task that includes encoding input data for reliable and/or efficient transmission or storage (and/or corresponding decoding). For example, the task may be an audio compression task. The input may include audio data and the output may comprise compressed audio data. In another example, the input includes visual data (e.g. one or more images or videos), the output comprises compressed visual data, and the task is a visual data compression task. In another example, the task may comprise generating an embedding for input data (e.g. input audio or visual data). In some cases, the input includes audio data representing a spoken utterance and the task is a speech recognition task. The output may comprise a text output which is mapped to the spoken utterance. In some cases, the task comprises encry pting or decrypting input data. In some cases, the task comprises a microprocessor performance task, such as branch prediction or memory address translation.

[0165] In some implementations, the task is a generative task, and machine-learned model(s) 1 can be configured to output content generated in view of input(s) 2. For instance, input(s) 2 can be or otherwise represent data of one or more modalities that encodes context for generating additional content.

[0166] In some implementations, the task can be a text completion task. Machine- learned model(s) 1 can be configured to process input(s) 2 that represent textual data and to generate output(s) 3 that represent additional textual data that completes a textual sequence that includes input(s) 2. For instance, machine-learned model(s) 1 can be configured to generate output(s) 3 to complete a sentence, paragraph, or portion of text that follows from a portion of text represented by input(s) 2.

[0167] In some implementations, the task can be an instruction following task. Machine- learned model(s) 1 can be configured to process input(s) 2 that represent instructions to perform a function and to generate output(s) 3 that advance a goal of satisfying the instruction function (e.g., at least a step of a multi-step procedure to perform the function). Output(s) 3 can represent data of the same or of a different modality as input(s) 2. For instance, input(s) 2 can represent textual data (e.g., natural language instructions for a task to be performed) and machine-learned model(s) 1 can process input(s) 2 to generate output(s) 3 that represent textual data responsive to the instructions (e.g., natural language responses, programming language responses, machine language responses, etc.). Input(s) 2 can represent image data (e.g., image-based instructions for a task to be performed, optionally accompanied by textual instructions) and machine-learned model(s) 1 can process input(s) 2 to generate output(s) 3 that represent textual data responsive to the instructions (e.g., natural language responses, programming language responses, machine language responses, etc.). One or more output(s) 3 can be iteratively or recursively generated to sequentially process and accomplish steps toward accomplishing the requested functionality. For instance, an initial output can be executed by an external system or be processed by machine-learned model(s) 1 to complete an initial step of performing a function. Multiple steps can be performed, with a final output being obtained that is responsive to the initial instructions.

[0168] In some implementations, the task can be a question answering task. Machine- learned model(s) 1 can be configured to process input(s) 2 that represent a question to answer and to generate output(s) 3 that advance a goal of returning an answer to the question (e g., at least a step of a multi-step procedure to perform the function). Output(s) 3 can represent data of the same or of a different modality as input(s) 2. For instance, input(s) 2 can represent textual data (e.g., natural language instructions for a task to be performed) and machine- learned model(s) 1 can process input(s) 2 to generate output(s) 3 that represent textual data responsive to the question (e.g., natural language responses, programming language responses, machine language responses, etc.). Input(s) 2 can represent image data (e.g., image-based instructions for a task to be performed, optionally accompanied by textual instructions) and machine-learned model(s) 1 can process input(s) 2 to generate output(s) 3 that represent textual data responsive to the question (e.g., natural language responses, programming language responses, machine language responses, etc.). One or more output(s) 3 can be iteratively or recursively generated to sequentially process and accomplish steps toward answering the question. For instance, an initial output can be executed by an external system or be processed by machine-learned model(s) 1 to complete an initial step of obtaining an answer to the question (e.g., querying a database, performing a computation, executing a script, etc.). Multiple steps can be performed, with a final output being obtained that is responsive to the question. [0169] In some implementations, the task can be an image generation task. Machine- learned model(s) 1 can be configured to process input(s) 2 that represent context regarding a desired portion of image content. The context can include text data, image data, audio data, etc. Machine-learned model(s) 1 can be configured to generate output(s) 3 that represent image data that depicts imagery⁷ related to the context. For instance, machine-learned model(s) 1 can be configured to generate pixel data of an image. Values for channel (s) associated with the pixels in the pixel data can be selected based on the context (e.g.. based on a probability determined based on the context).

[0170] In some implementations, the task can be an audio generation task. Machine- learned model(s) 1 can be configured to process input(s) 2 that represent context regarding a desired portion of audio content. The context can include text data, image data, audio data, etc. Machine-learned model(s) 1 can be configured to generate output(s) 3 that represent audio data related to the context. For instance, machine-learned model(s) 1 can be configured to generate waveform data in the form of an image (e.g., a spectrogram). Values for channel(s) associated with pixels of the image can be selected based on the context. Machine- learned model(s) 1 can be configured to generate waveform data in the form of a sequence of discrete samples of a continuous waveform. Values of the sequence can be selected based on the context (e.g., based on a probability⁷ determined based on the context).

[0171] In some implementations, the task can be a data generation task. Machine-learned model(s) 1 can be configured to process input(s) 2 that represent context regarding a desired portion of data (e.g., data from various data domains, such as sensor data, image data, multimodal data, statistical data, etc.). The desired data can be, for instance, synthetic data for training other machine-learned models. The context can include arbitrary data type(s). Machine-learned model(s) 1 can be configured to generate output(s) 3 that represent data that aligns with the desired data. For instance, machine-learned model (s) 1 can be configured to generate data values for populating a dataset. Values for the data object(s) can be selected based on the context (e.g., based on a probability⁷ determined based on the context).

Example Computing Systems and Devices

[0172] Figure 12 is a block diagram of an example networked computing system that can perform aspects of example implementations of the present disclosure. The system can include a number of computing devices and systems that are communicatively coupled over a network 49. An example computing device 50 is described to provide an example of a computing device that can perform any aspect of the present disclosure (e.g., implementing model host 31, client(s) 32, or both). An example server computing system 60 is described as an example of a server computing system that can perform any aspect of the present disclosure (e.g., implementing model host 31, client(s) 32, or both). Computing device 50 and server computing system(s) 60 can cooperatively interact (e.g., over network 49) to perform any aspect of the present disclosure (e.g., implementing model host 31, client(s) 32, or both). Model development platform system 70 is an example system that can host or serve model development platform(s) 12 for development of machine-learned models. Third-party system(s) 80 are example system(s) with which any of computing device 50, server computing system(s) 60, or model development platform system(s) 70 can interact in the performance of various aspects of the present disclosure (e.g., engaging third-party tools, accessing third-party databases or other resources, etc.).

[0173] Network 49 can be any type of communications network, such as a local area network (e.g., intranet), wide area network (e.g., Internet), or some combination thereof and can include any number of wired or wireless links. In general, communication over network 49 can be carried via any type of wired or wireless connection, using a wide variety of communication protocols (e.g., TCP/IP. HTTP. SMTP. FTP), encodings or formats (e.g., HTML, XML), or protection schemes (e.g., VPN, secure HTTP, SSL). Network 49 can also be implemented via a system bus. For instance, one or more devices or systems of Figure 12 can be co-located with, contained by, or otherwise integrated into one or more other devices or systems.

[0174] Computing device 50 can be any type of computing device, such as, for example, a personal computing device (e.g., laptop or desktop), a mobile computing device (e.g., smartphone or tablet), a gaming console or controller, a wearable computing device, an embedded computing device, a server computing device, a virtual machine operating on a host device, or any other type of computing device. Computing device 50 can be a client computing device. Computing device 50 can be an end-user computing device. Computing device 50 can be a computing device of a service provided that provides a service to an end user (who may use another computing device to interact with computing device 50).

[0175] Computing device 50 can include one or more processors 51 and a memory 52. Processor(s) 51 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected. Memory' 52 can include one or more non-transitory computer-readable storage media, such as HBM, RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. Memory 52 can store data 53 and instructions 54 which can be executed by processors ) 51 to cause computing device 50 to perform operations. The operations can implement any one or multiple features described herein. The operations can implement example methods and techniques described herein.

[0176] Computing device 50 can also include one or more input components that receive user input. For example, a user input component can be a touch-sensitive component (e.g., a touch-sensitive display screen or a touch pad) that is sensitive to the touch of a user input object (e.g., a finger or a stylus). The touch-sensitive component can serve to implement a virtual keyboard. Other example user input components include a microphone, camera, LIDAR, a physical keyboard or other buttons, or other means by which a user can provide user input.

[0177] Computing device 50 can store or include one or more machine-learned models 55. Machine-learned models 55 can include one or more machine-learned model(s) 1, such as a sequence processing model 4. Machine-learned models 55 can include one or multiple model instance(s) 31-1. Machine-learned model(s) 55 can be received from server computing system(s) 60. model development platform system 70, third party system(s) 80 (e.g.. an application distribution platform), or developed locally on computing device 50. Machine- learned model(s) 55 can be loaded into memory' 52 and used or otherwise implemented by processor(s) 51. Computing device 50 can implement multiple parallel instances of machine- learned model(s) 55.

[0178] Server computing system(s) 60 can include one or more processors 61 and a memory 62. Processor(s) 61 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA. a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected. Memory’ 62 can include one or more non-transitory computer-readable storage media, such as HBM, RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. Memory’ 62 can store data 63 and instructions 64 which can be executed by processor(s) 61 to cause server computing system(s) 60 to perform operations. The operations can implement any one or multiple features described herein. The operations can implement example methods and techniques described herein.

[0179] In some implementations, server computing system 60 includes or is otherwise implemented by one or multiple server computing devices. In instances in which server computing system 60 includes multiple server computing devices, such server computing devices can operate according to sequential computing architectures, parallel computing architectures, or some combination thereof.

[0180] Server computing system 60 can store or otherwise include one or more machine- learned models 65. Machine-learned model(s) 65 can be the same as or different from machine-learned model(s) 55. Machine-learned models 65 can include one or more machine- learned model(s) 1, such as a sequence processing model 4. Machine-learned models 65 can include one or multiple model instance(s) 31-1. Machine-learned model(s) 65 can be received from computing device 50, model development platform system 70, third party system(s) 80, or developed locally on server computing system(s) 60. Machine-learned model(s) 65 can be loaded into memory 62 and used or otherwise implemented by processor(s) 61. Server computing system(s) 60 can implement multiple parallel instances of machine-learned model(s) 65.

[0181] In an example configuration, machine-learned models 65 can be included in or otherwise stored and implemented by server computing system 60 to establish a client-server relationship with computing device 50 for serving model inferences. For instance, server computing system(s) 60 can implement model host 31 on behalf of client(s) 32 on computing device 50. For instance, machine-learned models 65 can be implemented by server computing system 60 as a portion of a w eb service (e.g., remote machine-learned model hosting sendee, such as an online interface for performing machine-learned model operations over a network on server computing system(s) 60). For instance, server computing system(s) 60 can communicate with computing device 50 over a local intranet or internet connection. For instance, computing device 50 can be a workstation or endpoint in communication with serv er computing system(s) 60, with implementation of machine-learned models 65 being managed by server computing system(s) 60 to remotely perform inference (e.g., for runtime or training operations), with output(s) returned (e.g., cast, streamed, etc.) to computing device 50. Machine-learned models 65 can w ork cooperatively or interoperatively with machine- learned models 55 on computing device 50 to perform various tasks.

[0182] In some implementations, the computing device 50 can store or include some or all of a text generation application 56. Alternatively or additionally, the server computing system 60 can store or include some or all of a text generation application 66. Thus, the text generation application can be located at the computing device 50 and/or at the server computing system 60. The text generation application 56 and/or 66 can be implemented in hardware, firmware, and/or software controlling a general purpose processor. For example, in some implementations, the text generation application 56 and/or 66 includes program files stored on a storage device, loaded into a memory and executed by one or more processors. In other implementations, the text generation application 56 and/or 66 includes one or more sets of computer-executable instructions that are stored in a tangible computer-readable storage medium such as RAM, hard disk, or optical or magnetic media.

[0183] Model development platform system(s) 70 can include one or more processors 71 and a memory 72. Processor(s) 71 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA. a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected. Memory 72 can include one or more non-transitory computer-readable storage media, such as HBM, RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. Memory 72 can store data 73 and instructions 74 which can be executed by processor(s) 71 to cause model development platform system(s) 70 to perform operations. The operations can implement any one or multiple features described herein. The operations can implement example methods and techniques described herein. Example operations include the functionality described herein with respect to model development platform 12. This and other functionality can be implemented by developer tool(s) 75.

[0184] Third-party system(s) 80 can include one or more processors 81 and a memory 82. Processor(s) 81 can be any suitable processing device (e.g., a processor core, a microprocessor, an ASIC, an FPGA, a controller, a microcontroller, etc.) and can be one processor or a plurality of processors that are operatively connected. Memory 82 can include one or more non-transitory computer-readable storage media, such as HBM, RAM, ROM, EEPROM, EPROM, flash memory devices, magnetic disks, etc., and combinations thereof. Memory 82 can store data 83 and instructions 84 which can be executed by processor(s) 81 to cause third-party system(s) 80 to perform operations. The operations can implement any one or multiple features described herein. The operations can implement example methods and techniques described herein. Example operations include the functionality described herein with respect to tools and other external resources called when training or performing inference with machine-learned model(s) 1, 4. 16. 20, 55, 65, etc. (e.g., third-party resource(s) 85).

[0185] Figure 12illustrates one example arrangement of computing systems that can be used to implement the present disclosure. Other computing system configurations can be used as well. For example, in some implementations, one or both of computing system 50 or server computing system(s) 60 can implement all or a portion of the operations of model development platform system 70. For example, computing system 50 or server computing system(s) 60 can implement developer tool(s) 75 (or extensions thereof) to develop, update/train, or refine machine-learned models 1, 4. 16. 20, 55, 65, etc. using one or more techniques described herein with respect to model alignment toolkit 17. In this manner, for instance, computing system 50 or server computing system(s) 60 can develop, update/train, or refine machine-learned models based on local datasets (e.g., for model personalization/customization. as permitted by user data preference selections).

[0186] Figure 13 is a block diagram of an example computing device 98 that performs according to example embodiments of the present disclosure. Computing device 98 can be a user computing device or a server computing device (e.g., computing device 50, server computing system(s) 60, etc.). Computing device 98 can implement model host 31. For instance, computing device 98 can include a number of applications (e.g., applications 1 through N). Each application can contain its own machine learning library and machine- learned model(s). For example, each application can include a machine-learned model. Example applications include a text messaging application, an email application, a dictation application, a virtual keyboard application, a browser application, etc. As illustrated in Figure 13. each application can communicate with a number of other components of the computing device, such as, for example, one or more sensors, a context manager, a device state component, or additional components. In some implementations, each application can communicate with each device component using an API (e.g., a public API). In some implementations, the API used by each application is specific to that application.

[0187] Figure 14 is a block diagram of an example computing device 99 that performs according to example embodiments of the present disclosure. Computing device 99 can be the same as or different from computing device 98. Computing device 99 can be a user computing device or a server computing device (e.g., computing device 50, server computing system(s) 60, etc.). Computing device 98 can implement model host 31. For instance, computing device 99 can include a number of applications (e.g., applications 1 through N). Each application can be in communication with a central intelligence layer. Example applications include a text messaging application, an email application, a dictation application, a virtual keyboard application, a browser application, etc. In some implementations, each application can communicate with the central intelligence layer (and model(s) stored therein) using an API (e.g., a common API across all applications).

[0188] The central intelligence layer can include a number of machine-learned models. For example, as illustrated in Figure 14. a respective machine-learned model can be provided for each application and managed by the central intelligence layer. In other implementations. two or more applications can share a single machine-learned model. For example, in some implementations, the central intelligence layer can provide a single model for all of the applications. In some implementations, the central intelligence layer is included within or otherwise implemented by an operating system of computing device 99.

[0189] The central intelligence layer can communicate with a central device data layer. The central device data layer can be a centralized repository’ of data for computing device 99. As illustrated in Figure 14. the central device data layer can communicate with a number of other components of the computing device, such as, for example, one or more sensors, a context manager, a device state component, or additional components. In some implementations, the central device data layer can communicate with each device component using an API (e.g.. a private API).

Additional Disclosure

[0190] The technology discussed herein makes reference to servers, databases, software applications, and other computer-based systems, as well as actions taken and information sent to and from such systems. The inherent flexibility of computer-based systems allows for a great variety of possible configurations, combinations, and divisions of tasks and functionality⁷ between and among components. For instance, processes discussed herein can be implemented using a single device or component or multiple devices or components working in combination. Databases and applications can be implemented on a single system or distributed across multiple systems. Distributed components can operate sequentially or in parallel.

[0191] While the present subject matter has been described in detail with respect to various specific example embodiments thereof, each example is provided by way of explanation, not limitation of the disclosure. Those skilled in the art, upon attaining an understanding of the foregoing, can readily produce alterations to, variations of, and equivalents to such embodiments. Accordingly, the subject disclosure does not preclude inclusion of such modifications, variations or additions to the present subject matter as would be readily apparent to one of ordinary skill in the art. For instance, features illustrated or described as part of one embodiment can be used with another embodiment to yield a still further embodiment. Thus, it is intended that the present disclosure cover such alterations, variations, and equivalents.

[0192] Aspects of the disclosure have been described in terms of illustrative embodiments thereof. Any and all features in the following claims can be combined or rearranged in any way possible, including combinations of claims not explicitly enumerated in combination together, as the example claim dependencies listed herein should not be read as limiting the scope of possible combinations of features disclosed herein. Accordingly, the scope of the present disclosure is by way of example rather than by way of limitation, and the subject disclosure does not preclude inclusion of such modifications, variations or additions to the present subject matter as would be readily apparent to one of ordinary skill in the art. Moreover, terms are described herein using lists of example elements joined by conjunctions such as “and,” “or,” “but,” etc. It should be understood that such conjunctions are provided for explanatory' purposes only. Clauses and other sequences of items joined by a particular conjunction such as “or,” for example, can refer to “and/or,” “at least one of”, “any combination of’ example elements listed therein, etc. Terms such as “based on” should be understood as “based at least in part on.”

[0193] The term “can” should be understood as referring to a possibility of a feature in various implementations and not as prescribing an ability that is necessarily present in every implementation. For example, the phrase “X can perform Y” should be understood as indicating that, in various implementations, X has the potential to be configured to perform Y, and not as indicating that in every instance X must always be able to perform Y. It should be understood that, in various implementations, X might be unable to perform Y and remain within the scope of the present disclosure.

[0194] The term “may” should be understood as referring to a possibility of a feature in various implementations and not as prescribing an ability that is necessarily present in every implementation. For example, the phrase “X may perform Y” should be understood as indicating that, in various implementations, X has the potential to be configured to perform Y, and not as indicating that in every’ instance X must always be able to perform Y. It should be understood that, in various implementations, X might be unable to perform Y and remain w ithin the scope of the present disclosure.

Claims

WHAT IS CLAIMED IS:

1. A computing system for automated text editing, the computing system comprising: a machine learning agent comprising one or more machine-learned models configured to perform one or more text editing tasks; and a text generation application, wherein the machine learning agent comprises an addressable entity that is addressable within the text generation application; wherein the computing system is configured to perform operations, the operations comprising: obtaining, by the computing system, data descriptive of an annotation input by a user in the text generation application, wherein the annotation corresponds to a set of textual content created within the text generation application, and wherein the annotation comprises a natural language text editing request that is addressed to the machine learning agent; in response to the annotation, executing, by the computing system, the machine learning agent to cause the machine learning agent to generate one or more edits to the set of textual content based on the natural language text editing request; and updating, by the computing system, the set of textual content based on the one or more edits generated by the machine learning agent.

2. The computing system of any preceding claim, wherein the annotation comprises a comment displayed in a margin of a graphical user interface of the text generation application.

3. The computing system of any preceding claim, wherein the annotation is addressed to the machine learning agent using natural language.

4. The computing system of claim 1 or claim 2, wherein the annotation is addressed to the machine learning agent using a particular symbol that designates the machine learning agent a recipient of the annotation.

5. The computing system of any preceding claim, wherein the set of textual content comprises two or more sentences.

6. The computing system of any preceding claim, wherein the set of textual content comprises an entirety of a document being edited by the user in the text generation application.

7. The computing system of any of claims 1-5, wherein the set of textual content comprises a subset of a document being edited by the user in the text generation application, and wherein the set of textual content has been selected by the user.

8. The computing system of any preceding claim, wherein executing, by the computing system, the machine learning agent comprises: generating, by the computing system, a model prompt based on the set of textual content and the natural language text editing request; and processing, by the computing system, the model prompt with the one or more machine-learned models of the machine learning agent to generate the one or more edits as an output of the one or more machine-learned models.

9. The computing system of claim 7, wherein generating, by the computing system, the model prompt based on the set of textual content and the natural language text editing request comprising: combining, by the computing system, (1) the set of textual content, (2) one or more context portions that are adjacent to the set of textual content within a document being edited by the user in the text generation application, and (3) the natural language text editing request to form the model prompt.

10. The computing system of any preceding claim, wherein the operations further comprise: executing, by the computing system, the machine learning agent to cause the machine learning agent to generate a second annotation that is responsive to the annotation input by the user; and inserting, by the computing system, the second annotation into the text generation application.

11. A computer-implemented method to train a language model, the method comprising: obtaining, by a computing system comprising one or more computing devices, a training tuple comprising a natural language text editing instruction, a set of textual content, and data describing one or more ground-truth edits to the set of textual content performed in associated with the natural language text editing instruction, wherein the set of textual content comprises two or more sentences; processing, by the computing system, the natural language text editing instruction and the set of textual content with the language model to generate one or more predicted edits to the set of textual content that are responsive to the natural language text editing instruction; evaluating, by the computing system, a loss function that generates a loss value based on a comparison of the one or more predicted edits to the one or more ground-truth edits; and modifying, by the computing system, one or more parameter values of the language model based on the loss function.

12. The computer-implemented method of claim 11, wherein the training tuple comprises a comment and edit history obtained from a text generation application in which the language model is deployed.

13. The computer-implemented method of claim 11, wherein the training tuple comprises an edit history obtained from a large text edit corpus.

14. The computer-implemented method of claim 13, wherein the large text edit corpus comprises edits to an online collaboratively-edited encyclopedia.

15. The computer-implemented method of any of claims 11-14, wherein the natural language instruction comprises a summary of the one or more ground-truth edits.

16. The computer-implemented method of any of claims 11-14, wherein the natural language instruction comprises a modified version of a summary of the one or more groundtruth edits, wherein the modified version of the summary was produced by a second, different language model from an original version of the summary of the one or more ground-truth edits.

17. The computer-implemented method of any of claims 11-16, wherein the natural language instruction has been selected by a reward language model.

18. The computer-implemented method of any of claims 11-17, wherein the natural language instruction has been filtered based on an NLI score, length, or an edit distance.

19. The computer-implemented method of any of claims 11-18, wherein the natural language instruction comprises a synthetic prompt.

20. One or more non-transitory computer readable media that collectively store the machine learning agent described in any of claims 1-10 or the language model described in any of claims 11-19.