KSNT News. We tell local Topeka and Northeast Kansas News & Weather stories. Local News That Matters.

AutoRABIT Receives Massive Security Boost with Addition of Cybersecurity 
Expert Jason Lord as Chief Information Security Officer

The research explores how consumers truly feel about their online shopping experiences. The report... AutoRABIT, the Salesforce DevOps Solution vendor, has launched LearningHub. This is the latest iteration of the AutoRABIT Academy. The AutoRABIT academy ...

butterflies-1127666_1280

Ashiss Kumar Dash

SAN FRANCISCO, Oct. 3, 2023 /PRNewswire/ -- AutoRABIT has prioritized improving the quality of its products and enhancing the user experience. The introduction of LearningHub?a more mature and ...

AutoRABIT Introduces LearningHub to Improve the User Experience

The new platform is a comprehensive resource for learning and collaboration. SAN FRANCISCO, Oct. 3, 2023 /PRNewswire/ -- AutoRABIT has prioritized improving the quality of its products and enhancing the user experience. The introduction of LearningHub—a ...

SAN FRANCISCO, Oct. 3, 2023 /PRNewswire/ -- AutoRABIT has prioritized improving the quality of its products and enhancing the user experience. The introduction of LearningHub—a more mature and ...

AutoRABIT, the Salesforce DevOps Solution vendor, has launched LearningHub. This is the latest iteration of the AutoRABIT Academy. The AutoRABIT academy will be available until November 30 th, when the site will re-direct to the LearningHub. The ...

AutoRABIT LearningHUB, a rolling snowball or stone?

AutoRABIT

autorabit

With digital transformation and the move to online operations, companies are increasingly relying on Salesforce to manage many of their operations.  In this context, teams are turning to Salesforce CI/CD to automate manual workflows, which help them handle these growing workloads and quickly deliver quality work to their end-users. This article gives you a short overview of how to implement CI/CD in Salesforce DX. Components of CI/CD A development process that uses version control as the source for new development activity is used to build Salesforce CI/CD . The way you manage and coordinate concurrent work streams and multiple environments are then determined by the Git branching strategy you choose. Automation is used in this process to validate deployment packages, run tests, and move work items along the pipeline. Individual CI jobs can be set up using a solution such as AutoRABIT, Gearset or a standard automation tool such as Jenkins. The continuous integration in DevOps processes runs on a regular basis or is triggered when branches in Git merge, and they can conduct tests before deployments across version control and Salesforce environments. Salesforce DevOps Implementation of CI/CD with Salesforce DX Implementation of the Salesforce CI/CD pipeline is separated into four stages: Source Build Test Deployment Before proceeding to the next stage, each succeeding stage must be accomplished. All steps are constantly checked for faults or anomalies, and feedback is sent to the delivery team. Each development, whether a bug patch or a feature enhancement, is routed via the CI/CD pipeline in DevOps before being deployed to production. Source It is the first stage of Salesforce CI/CD workflow. Here the CI/CD pipeline in DevOps will be activated by any change in the program or a predefined flag in the code repository. This level is all about source control, including version control and change tracking. When the automated process detects a change in the central repository (commit, new version), it will start tasks like code compilation and unit testing. Build This stage of the pipeline converts the source code and all of its dependencies into an executable/runnable instance of the development. This stage addresses software builds as well as other types of buildable objects such as Docker containers. Failure in this build could suggest a fundamental problem with the underlying code. A centralized artifact repository such as yarn, JFrog, or a cloud-based solution such as Azure Artifacts can be used to centralize the storage process of building artifacts. If there are any issues with the current build, we will be able to roll back to the previous build. Test The test stage includes all automated testing to validate the software's behavior. This stage's purpose is to keep software defects from reaching end users. This stage might include a variety of methods of testing, ranging from integration testing to functional testing. This stage will also reveal any flaws in the product. Deploy This is the Salesforce CI/CD pipeline's final stage. The package is now ready to be deployed after passing all of the preceding phases. At this point, the package is deployed to appropriate environments, first to a staging environment for additional quality assurance (QA) and then to a production environment. Infrastructure provisioning, configuration, and containerization utilizing technologies such as Terraform, Puppet, Docker, and Kubernetes can all be part of the deployment stage. Conclusion New business needs and opportunities constantly increase the workload of Salesforce teams. So, teams are turning to Salesforce CI/CD to automate manual workflows, which help them handle these growing workloads and quickly deliver quality work to their end-users

CI/CD Best Practices for Salesforce DX

Meet the Winners of Startups in North and Central America

Analyzing the Impact of Model Scaling on Long-Form Factuality

How AI Breaks Down and Validates Information for Truthfulness

Table of Links Abstract and 1 Introduction 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 3 Safe:LLM agents as factuality autoraters 4 LLMs agents can be better factuality annotators than humans 5 F1@k: Extending F1 with recall from human-preferred length 6 Larger LLMs are more factual 7 Related Work 8 Limitations 9 Conclusion, Acknowledgments, Author Contribution, and References Appendix A. Frequently asked questions B. LongFact details C. SAFE details D. Metric details E. Further analysis B LONGFACT DETAILS B.1 DATA-GENERATION PROCESS To create LongFact, we first manually selected a set of 38 topics (shown in Appendix B.2) that provide broad coverage of long-form factuality. For a given topic, we generate prompts by instructing GPT-4[16] to come up with questions that require long-form responses and that are about specific and niche concepts (LongFact-Concepts) or objects (LongFact-Objects) within that topic. We generate prompts one at a time, obtaining one new prompt after each model call. During each model call, we provide a set of ten in-context exemplars, which starts as a set of ten manually-created exemplars (shown in Table 3) but gradually gets replaced with generated prompts as each model call generates a new prompt (a changing set of in-context exemplars helps increase prompt diversity). The selected ten in-context exemplars and the given topic are inputted into the prompt templates shown in Table 4 and passed to the model to generate the next prompt.[17] We continue this process until the desired number of prompts are generated. Source code for generating LongFact using this process can be found at https://github.com/google-deepmind/long-form-factuality. We first use this process to generate 60 prompts per topic for both LongFact-Concepts and LongFactObjects at a total cost of approximately $260 in OpenAI API credits. For each topic, we then manually remove any duplicated questions, which we define as any two questions that can be answered with approximately the same response. To balance the number of prompts per topic, we randomly select 30 prompts to keep from the set of deduplicated questions. This results in a total of 1,140 prompts for LongFact-Concepts and 1,140 prompts for LongFact-Objects. We release both LongFact-Concepts and LongFact-Objects at https://github.com/google-deepmind/ long-form-factuality/tree/main/longfact. B.2 LIST OF TOPICS We manually selected a set of 38 topics for LongFact in order to provide broad coverage of long-form factuality across many topics. LongFact includes topics from MMLU (Hendrycks et al., 2021) for broader topic coverage, as well as topics such as celebrities, gaming, movies, and music, which we added in an effort to account for topics that are less academic in nature. Table 6 shows each topic, a list of example concepts and objects within that topic that are tested by LongFact, and a classification of that topic into a supercategory of either “STEM”, “social sciences”, “humanities”, or “other” (see Figure 2 for the ratio of supercategories across all topics). Example prompts from LongFact-Concepts and LongFact-Objects for each topic are shown in Appendix B.3. B.3 EXAMPLE PROMPTS This paper is available on arxiv under CC by 4.0 Deed license. [16] We use gpt-4-0613 at a temperature of 1.0 and a max decode length of 128 tokens. [17] The “moral disputes” topic in LongFact-Objects requires a custom set of instructions to obtain prompts that ask about specific objects (see Table 5). The in-context exemplars remain the same. Authors:
(1) Jerry Wei, Google DeepMind and a Lead contributors;
(2) Chengrun Yang, Google DeepMind and a Lead contributors;
(3) Xinying Song, Google DeepMind and a Lead contributors;
(4) Yifeng Lu, Google DeepMind and a Lead contributors;
(5) Nathan Hu, Google DeepMind and Stanford University;
(6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign;
(7) Dustin Tran, Google DeepMind;
(8) Daiyi Peng, Google DeepMind;
(9) Ruibo Liu, Google DeepMind;
(10) Da Huang, Google DeepMind;
(11) Cosmo Du, Google DeepMind;
(12) Quoc V. Le, Google DeepMind. Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 3 Safe:LLM agents as factuality autoraters 3 Safe:LLM agents as factuality autoraters 4 LLMs agents can be better factuality annotators than humans 4 LLMs agents can be better factuality annotators than humans 5 F1@k: Extending F1 with recall from human-preferred length 5 F1@k: Extending F1 with recall from human-preferred length 6 Larger LLMs are more factual 6 Larger LLMs are more factual 7 Related Work 7 Related Work 8 Limitations 8 Limitations 9 Conclusion, Acknowledgments, Author Contribution, and References 9 Conclusion, Acknowledgments, Author Contribution, and References Appendix Appendix A. Frequently asked questions A. Frequently asked questions B. LongFact details B. LongFact details C. SAFE details C. SAFE details D. Metric details D. Metric details E. Further analysis E. Further analysis B LONGFACT DETAILS B.1 DATA-GENERATION PROCESS To create LongFact, we first manually selected a set of 38 topics (shown in Appendix B.2) that provide broad coverage of long-form factuality. For a given topic, we generate prompts by instructing GPT-4[16] to come up with questions that require long-form responses and that are about specific and niche concepts (LongFact-Concepts) or objects (LongFact-Objects) within that topic. We generate prompts one at a time, obtaining one new prompt after each model call. During each model call, we provide a set of ten in-context exemplars, which starts as a set of ten manually-created exemplars (shown in Table 3) but gradually gets replaced with generated prompts as each model call generates a new prompt (a changing set of in-context exemplars helps increase prompt diversity). The selected ten in-context exemplars and the given topic are inputted into the prompt templates shown in Table 4 and passed to the model to generate the next prompt.[17] We continue this process until the desired number of prompts are generated. Source code for generating LongFact using this process can be found at https://github.com/google-deepmind/long-form-factuality. We first use this process to generate 60 prompts per topic for both LongFact-Concepts and LongFactObjects at a total cost of approximately $260 in OpenAI API credits. For each topic, we then manually remove any duplicated questions, which we define as any two questions that can be answered with approximately the same response. To balance the number of prompts per topic, we randomly select 30 prompts to keep from the set of deduplicated questions. This results in a total of 1,140 prompts for LongFact-Concepts and 1,140 prompts for LongFact-Objects. We release both LongFact-Concepts and LongFact-Objects at https://github.com/google-deepmind/ long-form-factuality/tree/main/longfact. B.2 LIST OF TOPICS We manually selected a set of 38 topics for LongFact in order to provide broad coverage of long-form factuality across many topics. LongFact includes topics from MMLU (Hendrycks et al., 2021) for broader topic coverage, as well as topics such as celebrities, gaming, movies, and music, which we added in an effort to account for topics that are less academic in nature. Table 6 shows each topic, a list of example concepts and objects within that topic that are tested by LongFact, and a classification of that topic into a supercategory of either “STEM”, “social sciences”, “humanities”, or “other” (see Figure 2 for the ratio of supercategories across all topics). Example prompts from LongFact-Concepts and LongFact-Objects for each topic are shown in Appendix B.3. B.3 EXAMPLE PROMPTS This paper is available on arxiv under CC by 4.0 Deed license. This paper is available on arxiv under CC by 4.0 Deed license. available on arxiv available on arxiv [16] We use gpt-4-0613 at a temperature of 1.0 and a max decode length of 128 tokens. [17] The “moral disputes” topic in LongFact-Objects requires a custom set of instructions to obtain prompts that ask about specific objects (see Table 5). The in-context exemplars remain the same. Authors: (1) Jerry Wei, Google DeepMind and a Lead contributors; (2) Chengrun Yang, Google DeepMind and a Lead contributors; (3) Xinying Song, Google DeepMind and a Lead contributors; (4) Yifeng Lu, Google DeepMind and a Lead contributors; (5) Nathan Hu, Google DeepMind and Stanford University; (6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign; (7) Dustin Tran, Google DeepMind; (8) Daiyi Peng, Google DeepMind; (9) Ruibo Liu, Google DeepMind; (10) Da Huang, Google DeepMind; (11) Cosmo Du, Google DeepMind; (12) Quoc V. Le, Google DeepMind. Authors: Authors: (1) Jerry Wei, Google DeepMind and a Lead contributors; (2) Chengrun Yang, Google DeepMind and a Lead contributors; (3) Xinying Song, Google DeepMind and a Lead contributors; (4) Yifeng Lu, Google DeepMind and a Lead contributors; (5) Nathan Hu, Google DeepMind and Stanford University; (6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign; (7) Dustin Tran, Google DeepMind; (8) Daiyi Peng, Google DeepMind; (9) Ruibo Liu, Google DeepMind; (10) Da Huang, Google DeepMind; (11) Cosmo Du, Google DeepMind; (12) Quoc V. Le, Google DeepMind.

How LongFact Helps Measure the Accuracy of AI Responses

How SAFE Performs Compared to Human Annotations

Benchmarking Long-Form Factuality in Large Language Models

Table of Links Abstract and 1 Introduction 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 3 Safe:LLM agents as factuality autoraters 4 LLMs agents can be better factuality annotators than humans 5 F1@k: Extending F1 with recall from human-preferred length 6 Larger LLMs are more factual 7 Related Work 8 Limitations 9 Conclusion, Acknowledgments, Author Contribution, and References Appendix A. Frequently asked questions B. LongFact details C. SAFE details D. Metric details E. Further analysis 8 LIMITATIONS Both LongFact and SAFE rely on LLMs to function, hence the used-LLM’s capabilities (especially instruction following and reasoning, but also abilities such creativity) directly affect the quality of the generated LongFact prompts and SAFE. With respect to generating LongFact, a language model that cannot follow instructions may generate prompts that are not within the given topic or that do not solicit long-form responses. Since model outputs to generate LongFact are relatively short, however, we use GPT-4 to maximize the quality of the generated prompts. In SAFE, a weak language model may (a) decompose a long-form response into individual facts that contain more than one factual statement, (b) exhibit poor reasoning when determining the relevance of an individual fact to the prompt, (c) propose unrelated search queries to verify a fact, or (d) fail to reason about whether a fact is supported by search results. For this reason, our configuration of SAFE uses GPT-3.5-Turbo, which provides strong performance in most cases at low cost, though it still exhibits some failures due to model capability (further examined in Appendix A.3). Future work may examine whether finetuning a cost-effective language model to perform the steps in SAFE may reduce failures while maintaining a low inference-time cost. There also exists an inherent weakness in SAFE, which is that it relies on Google Search as a knowledge source to obtain ground-truths, which may not suffice in corner cases. For example, Google Search may not easily find information about some fact or may lack profundity in expert-level domains such as law and medicine. At the same time, however, Google Search allows access to the entire internet and thus is arguably the most-comprehensive knowledge source for open-domain factuality tasks that is also widely-accessible. We alleviate this concern by reporting our labels as “supported” or “not supported” by Google Search results, rather than attempting to label facts as globally factual or non-factual (for discussion on label selection, see Appendix A.2), though future work may investigate whether there are better proxies for global factuality than verification by Google Search. We discuss other limitations of and avenues of exploration for SAFE in Appendix C.6. Furthermore, while F1@K is a robust aggregation of factual precision and recall, it operates under the assumption that a response does not contain any repeated facts. While this is a reasonable assumption in practice, it also provides the opportunity for our metric to be gamed by repeating a supported fact. As stated in Appendix D.1, however, we contend that repetition of facts can be better measured by other metrics such as fluency or usefulness of model responses. If desired, we also suggest that a step of removing duplicated facts can be added into SAFE, though we leave this for future research due to the added complexity required to account for this case. This paper is available on arxiv under CC by 4.0 Deed license. Authors:
(1) Jerry Wei, Google DeepMind and a Lead contributors;
(2) Chengrun Yang, Google DeepMind and a Lead contributors;
(3) Xinying Song, Google DeepMind and a Lead contributors;
(4) Yifeng Lu, Google DeepMind and a Lead contributors;
(5) Nathan Hu, Google DeepMind and Stanford University;
(6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign;
(7) Dustin Tran, Google DeepMind;
(8) Daiyi Peng, Google DeepMind;
(9) Ruibo Liu, Google DeepMind;
(10) Da Huang, Google DeepMind;
(11) Cosmo Du, Google DeepMind;
(12) Quoc V. Le, Google DeepMind. Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 3 Safe:LLM agents as factuality autoraters 3 Safe:LLM agents as factuality autoraters 4 LLMs agents can be better factuality annotators than humans 4 LLMs agents can be better factuality annotators than humans 5 F1@k: Extending F1 with recall from human-preferred length 5 F1@k: Extending F1 with recall from human-preferred length 6 Larger LLMs are more factual 6 Larger LLMs are more factual 7 Related Work 7 Related Work 8 Limitations 8 Limitations 9 Conclusion, Acknowledgments, Author Contribution, and References 9 Conclusion, Acknowledgments, Author Contribution, and References Appendix Appendix A. Frequently asked questions A. Frequently asked questions B. LongFact details B. LongFact details C. SAFE details C. SAFE details D. Metric details D. Metric details E. Further analysis E. Further analysis 8 LIMITATIONS Both LongFact and SAFE rely on LLMs to function, hence the used-LLM’s capabilities (especially instruction following and reasoning, but also abilities such creativity) directly affect the quality of the generated LongFact prompts and SAFE. With respect to generating LongFact, a language model that cannot follow instructions may generate prompts that are not within the given topic or that do not solicit long-form responses. Since model outputs to generate LongFact are relatively short, however, we use GPT-4 to maximize the quality of the generated prompts. In SAFE, a weak language model may (a) decompose a long-form response into individual facts that contain more than one factual statement, (b) exhibit poor reasoning when determining the relevance of an individual fact to the prompt, (c) propose unrelated search queries to verify a fact, or (d) fail to reason about whether a fact is supported by search results. For this reason, our configuration of SAFE uses GPT-3.5-Turbo, which provides strong performance in most cases at low cost, though it still exhibits some failures due to model capability (further examined in Appendix A.3). Future work may examine whether finetuning a cost-effective language model to perform the steps in SAFE may reduce failures while maintaining a low inference-time cost. There also exists an inherent weakness in SAFE, which is that it relies on Google Search as a knowledge source to obtain ground-truths, which may not suffice in corner cases. For example, Google Search may not easily find information about some fact or may lack profundity in expert-level domains such as law and medicine. At the same time, however, Google Search allows access to the entire internet and thus is arguably the most-comprehensive knowledge source for open-domain factuality tasks that is also widely-accessible. We alleviate this concern by reporting our labels as “supported” or “not supported” by Google Search results, rather than attempting to label facts as globally factual or non-factual (for discussion on label selection, see Appendix A.2), though future work may investigate whether there are better proxies for global factuality than verification by Google Search. We discuss other limitations of and avenues of exploration for SAFE in Appendix C.6. Furthermore, while F1@K is a robust aggregation of factual precision and recall, it operates under the assumption that a response does not contain any repeated facts. While this is a reasonable assumption in practice, it also provides the opportunity for our metric to be gamed by repeating a supported fact. As stated in Appendix D.1, however, we contend that repetition of facts can be better measured by other metrics such as fluency or usefulness of model responses. If desired, we also suggest that a step of removing duplicated facts can be added into SAFE, though we leave this for future research due to the added complexity required to account for this case. This paper is available on arxiv under CC by 4.0 Deed license. This paper is available on arxiv under CC by 4.0 Deed license. available on arxiv available on arxiv Authors: (1) Jerry Wei, Google DeepMind and a Lead contributors; (2) Chengrun Yang, Google DeepMind and a Lead contributors; (3) Xinying Song, Google DeepMind and a Lead contributors; (4) Yifeng Lu, Google DeepMind and a Lead contributors; (5) Nathan Hu, Google DeepMind and Stanford University; (6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign; (7) Dustin Tran, Google DeepMind; (8) Daiyi Peng, Google DeepMind; (9) Ruibo Liu, Google DeepMind; (10) Da Huang, Google DeepMind; (11) Cosmo Du, Google DeepMind; (12) Quoc V. Le, Google DeepMind. Authors: Authors: (1) Jerry Wei, Google DeepMind and a Lead contributors; (2) Chengrun Yang, Google DeepMind and a Lead contributors; (3) Xinying Song, Google DeepMind and a Lead contributors; (4) Yifeng Lu, Google DeepMind and a Lead contributors; (5) Nathan Hu, Google DeepMind and Stanford University; (6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign; (7) Dustin Tran, Google DeepMind; (8) Daiyi Peng, Google DeepMind; (9) Ruibo Liu, Google DeepMind; (10) Da Huang, Google DeepMind; (11) Cosmo Du, Google DeepMind; (12) Quoc V. Le, Google DeepMind.

Challenges in Using Google Search for Factuality Verification

Table of Links Abstract and 1 Introduction 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 3 Safe:LLM agents as factuality autoraters 4 LLMs agents can be better factuality annotators than humans 5 F1@k: Extending F1 with recall from human-preferred length 6 Larger LLMs are more factual 7 Related Work 8 Limitations 9 Conclusion, Acknowledgments, Author Contribution, and References Appendix A. Frequently asked questions B. LongFact details C. SAFE details D. Metric details E. Further analysis 7 RELATED WORK Benchmarking factuality. Recent studies on factuality in large language models have presented several factuality benchmarks. Many benchmarks such as FELM (Chen et al., 2023), FreshQA (Vu et al., 2023), and HaluEval (Li et al., 2023) simply test a model’s knowledge of various factoids. On the other hand, adversarial benchmarks such as TruthfulQA (Lin et al., 2022) and HalluQA (Cheng et al., 2023b) consist of prompts that are designed to test whether a language model will generate a false factoid learned from imitating popular misconceptions. In the long-form setting, Min et al. (2023) proposed a set of prompts that require language models to generate biographies about people, evaluated against each person’s respective Wikipedia article. Our proposed prompt set does not dispute the importance of these tasks, but rather aims to provide broad coverage of factuality in a long-form setting, whereas existing benchmarks either cover singular short factoids or only cover a small set of long-form topics. Evaluating factuality in model responses. In order to fully benchmark a model’s factuality, there must exist a reliable method for quantifying the factuality of a model response to a given prompt. Performance on factuality datasets with short ground-truth answers can often be evaluated using language models that have been fine-tuned on human evaluations (Lin et al., 2022; Sellam et al., 2020; Lin, 2004). Long-form responses, on the other hand, are harder to evaluate because they contain a much-broader range of possible factual claims. For example, FActScore (Min et al., 2023) leverages Wikipedia pages as knowledge sources for humans and language models to use when rating long-form responses on biographies about people. In reference-free contexts, RAGAS (Es et al., 2023) evaluate whether a response from a retrieval-augmented generation (RAG) system is self-contained by prompting a large language model to check the consistency among questions, retrieved contexts, answers, and statements within the answer. This matches the case study in Guan et al. (2023), which also showed that language models may serve as a strong tool for fact verification. Tian et al. (2023) leverages this intuition to demonstrate that factuality preference rankings obtained via model confidence can be used for improving factuality in responses via direct preference optimization. Improvements in human evaluation can also be important for evaluating factuality; in this domain, Cheng et al. (2023a) created an interface for human evaluation by asking fact-verification questions and checking self-consistency of model responses. Most related to our work, Chern et al. (2023) proposed a method of rating model responses by breaking them down and rating individual components via external tools and applied this in tasks such as code generation, math reasoning, and short-answer QA. Our proposed method takes inspiration from this work, but applies it in the long-form factuality setting, similar to other work that uses search-augmented language models for fact verification (Gao et al., 2023; Wang et al., 2023; Zhang & Gao, 2023). Compared to these prior methods of evaluating factuality, SAFE is automatic, does not require any model finetuning or any preset reference answers/knowledge sources, and is more-reliable than crowdsourced human annotations while only requiring a fraction of the cost. Quantifying long-form factuality. Long-form factuality is difficult to quantify because the quality of the response is influenced both by the factuality of the response (precision) and by the coverage of the response (recall). FActScore (Min et al., 2023) measures precision with respect to Wikipedia pages as knowledge sources and mentions the difficulty of measuring recall as future work. Tian et al. (2023) and Dhuliawala et al. (2023) also use precision as a proxy for a model response’s factuality. Other metrics such as token-level F1 score (Rajpurkar et al., 2016), sentence or passage-level AUC (Manakul et al., 2023), and similarity to ground-truth (Wieting et al., 2022) can account for recall as well as precision but require human responses or judgements that serve as a ground-truth. Our proposed F1@K, on the other hand, measures both precision and recall of a long-form response without requiring any preset ground-truth answers, only the number of facts from each label category. This paper is available on arxiv under CC by 4.0 Deed license. Authors:
(1) Jerry Wei, Google DeepMind and a Lead contributors;
(2) Chengrun Yang, Google DeepMind and a Lead contributors;
(3) Xinying Song, Google DeepMind and a Lead contributors;
(4) Yifeng Lu, Google DeepMind and a Lead contributors;
(5) Nathan Hu, Google DeepMind and Stanford University;
(6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign;
(7) Dustin Tran, Google DeepMind;
(8) Daiyi Peng, Google DeepMind;
(9) Ruibo Liu, Google DeepMind;
(10) Da Huang, Google DeepMind;
(11) Cosmo Du, Google DeepMind;
(12) Quoc V. Le, Google DeepMind. Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 3 Safe:LLM agents as factuality autoraters 3 Safe:LLM agents as factuality autoraters 4 LLMs agents can be better factuality annotators than humans 4 LLMs agents can be better factuality annotators than humans 5 F1@k: Extending F1 with recall from human-preferred length 5 F1@k: Extending F1 with recall from human-preferred length 6 Larger LLMs are more factual 6 Larger LLMs are more factual 7 Related Work 7 Related Work 8 Limitations 8 Limitations 9 Conclusion, Acknowledgments, Author Contribution, and References 9 Conclusion, Acknowledgments, Author Contribution, and References Appendix Appendix A. Frequently asked questions A. Frequently asked questions B. LongFact details B. LongFact details C. SAFE details C. SAFE details D. Metric details D. Metric details E. Further analysis E. Further analysis 7 RELATED WORK Benchmarking factuality. Recent studies on factuality in large language models have presented several factuality benchmarks. Many benchmarks such as FELM (Chen et al., 2023), FreshQA (Vu et al., 2023), and HaluEval (Li et al., 2023) simply test a model’s knowledge of various factoids. On the other hand, adversarial benchmarks such as TruthfulQA (Lin et al., 2022) and HalluQA (Cheng et al., 2023b) consist of prompts that are designed to test whether a language model will generate a false factoid learned from imitating popular misconceptions. In the long-form setting, Min et al. (2023) proposed a set of prompts that require language models to generate biographies about people, evaluated against each person’s respective Wikipedia article. Our proposed prompt set does not dispute the importance of these tasks, but rather aims to provide broad coverage of factuality in a long-form setting, whereas existing benchmarks either cover singular short factoids or only cover a small set of long-form topics. Benchmarking factuality. Evaluating factuality in model responses. In order to fully benchmark a model’s factuality, there must exist a reliable method for quantifying the factuality of a model response to a given prompt. Performance on factuality datasets with short ground-truth answers can often be evaluated using language models that have been fine-tuned on human evaluations (Lin et al., 2022; Sellam et al., 2020; Lin, 2004). Long-form responses, on the other hand, are harder to evaluate because they contain a much-broader range of possible factual claims. For example, FActScore (Min et al., 2023) leverages Wikipedia pages as knowledge sources for humans and language models to use when rating long-form responses on biographies about people. In reference-free contexts, RAGAS (Es et al., 2023) evaluate whether a response from a retrieval-augmented generation (RAG) system is self-contained by prompting a large language model to check the consistency among questions, retrieved contexts, answers, and statements within the answer. This matches the case study in Guan et al. (2023), which also showed that language models may serve as a strong tool for fact verification. Tian et al. (2023) leverages this intuition to demonstrate that factuality preference rankings obtained via model confidence can be used for improving factuality in responses via direct preference optimization. Improvements in human evaluation can also be important for evaluating factuality; in this domain, Cheng et al. (2023a) created an interface for human evaluation by asking fact-verification questions and checking self-consistency of model responses. Most related to our work, Chern et al. (2023) proposed a method of rating model responses by breaking them down and rating individual components via external tools and applied this in tasks such as code generation, math reasoning, and short-answer QA. Our proposed method takes inspiration from this work, but applies it in the long-form factuality setting, similar to other work that uses search-augmented language models for fact verification (Gao et al., 2023; Wang et al., 2023; Zhang & Gao, 2023). Compared to these prior methods of evaluating factuality, SAFE is automatic, does not require any model finetuning or any preset reference answers/knowledge sources, and is more-reliable than crowdsourced human annotations while only requiring a fraction of the cost. Evaluating factuality in model responses. Quantifying long-form factuality. Long-form factuality is difficult to quantify because the quality of the response is influenced both by the factuality of the response (precision) and by the coverage of the response (recall). FActScore (Min et al., 2023) measures precision with respect to Wikipedia pages as knowledge sources and mentions the difficulty of measuring recall as future work. Tian et al. (2023) and Dhuliawala et al. (2023) also use precision as a proxy for a model response’s factuality. Other metrics such as token-level F1 score (Rajpurkar et al., 2016), sentence or passage-level AUC (Manakul et al., 2023), and similarity to ground-truth (Wieting et al., 2022) can account for recall as well as precision but require human responses or judgements that serve as a ground-truth. Our proposed F1@K, on the other hand, measures both precision and recall of a long-form response without requiring any preset ground-truth answers, only the number of facts from each label category. Quantifying long-form factuality. This paper is available on arxiv under CC by 4.0 Deed license. This paper is available on arxiv under CC by 4.0 Deed license. available on arxiv available on arxiv Authors: (1) Jerry Wei, Google DeepMind and a Lead contributors; (2) Chengrun Yang, Google DeepMind and a Lead contributors; (3) Xinying Song, Google DeepMind and a Lead contributors; (4) Yifeng Lu, Google DeepMind and a Lead contributors; (5) Nathan Hu, Google DeepMind and Stanford University; (6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign; (7) Dustin Tran, Google DeepMind; (8) Daiyi Peng, Google DeepMind; (9) Ruibo Liu, Google DeepMind; (10) Da Huang, Google DeepMind; (11) Cosmo Du, Google DeepMind; (12) Quoc V. Le, Google DeepMind. Authors: Authors: (1) Jerry Wei, Google DeepMind and a Lead contributors; (2) Chengrun Yang, Google DeepMind and a Lead contributors; (3) Xinying Song, Google DeepMind and a Lead contributors; (4) Yifeng Lu, Google DeepMind and a Lead contributors; (5) Nathan Hu, Google DeepMind and Stanford University; (6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign; (7) Dustin Tran, Google DeepMind; (8) Daiyi Peng, Google DeepMind; (9) Ruibo Liu, Google DeepMind; (10) Da Huang, Google DeepMind; (11) Cosmo Du, Google DeepMind; (12) Quoc V. Le, Google DeepMind.

A Smarter Way to Check If AI Answers Are Correct

Table of Links Abstract and 1 Introduction Abstract and 1 Introduction 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 2 LongFact: Using LLMs to generate a multi-topic benchmark for long-form factuality 3 Safe:LLM agents as factuality autoraters 3 Safe:LLM agents as factuality autoraters 4 LLMs agents can be better factuality annotators than humans 4 LLMs agents can be better factuality annotators than humans 5 F1@k: Extending F1 with recall from human-preferred length 5 F1@k: Extending F1 with recall from human-preferred length 6 Larger LLMs are more factual 6 Larger LLMs are more factual 7 Related Work 7 Related Work 8 Limitations 8 Limitations 9 Conclusion, Acknowledgments, Author Contribution, and References 9 Conclusion, Acknowledgments, Author Contribution, and References Appendix Appendix A. Frequently asked questions A. Frequently asked questions B. LongFact details B. LongFact details C. SAFE details C. SAFE details D. Metric details D. Metric details E. Further analysis E. Further analysis 4 LLM AGENTS CAN BE BETTER FACTUALITY ANNOTATORS THAN HUMANS LLMs have achieved superhuman performance on reasoning benchmarks (Bubeck et al., 2023; Gemini Team, 2023; Li et al., 2022) and higher factuality than humans on summarization tasks (Pu et al., 2023). The use of human annotators, however, is still prevalent in language-model research, which often creates a bottleneck due to human annotators’ high cost and variance in both evaluation speed and quality. Because of this mismatch, we investigate how SAFE compares to human annotations and whether it can replace human raters in evaluating long-form factuality To quantitatively evaluate the quality of annotations obtained using SAFE, we leverage crowdsourced human annotations released from Min et al. (2023). These data contain 496 prompt–response pairs where responses have been manually split into individual facts (for a total of 16,011 individual facts), and each individual fact has been manually labeled as either supported, irrelevant, or not supported. For each individual fact, we obtain an annotation from SAFE by running the SAFE pipeline, starting from revising the fact to be self-contained (i.e., steps 2–4 in Figure 1).[6] We then compare SAFE annotations with human annotations at the individual-fact level. First, we directly compare each individual fact’s SAFE annotation and human annotation, finding that SAFE agrees with humans on 72.0% of individual facts (see Figure 4). This indicates that SAFE achieves human-level performance on a majority of individual facts. We then examine a randomly-sampled subset of 100 individual facts for which the annotation from SAFE disagreed with the annotation from human raters. We manually re-annotate each of these facts (allowing access to Google Search rather than only Wikipedia for more-comprehensive annotations), and we use these labels as the ground-truth. We found that on these disagreement cases, SAFE annotations were correct 76% of the time, whereas human annotations were correct only 19% (see Figure 5), which represents a 4 to 1 win ratio for SAFE. Examples of correct SAFE ratings and incorrect SAFE ratings are shown in Appendix C.2 and Appendix C.3, respectively. Appendix A.3 analyzes failure causes for SAFE, and Appendix A.4 analyzes presumed failure causes for human annotators. To rate all individual facts from the 496 prompt–response pairs, SAFE issued GPT-3.5-Turbo API calls costing $64.57 and Serper API calls costing $31.74, thus costing a total of $96.31, equivalent to $0.19 per model response. For comparison, Min et al. (2023) reported a cost of $4 per model response to obtain crowdsourced human annotations for the dataset. This means that in addition to outperforming human raters, SAFE is also more than 20× cheaper than crowdsourced human annotators. Overall, the superior performance of SAFE demonstrates that language models can be used as scalable autoraters that have the potential to outperform crowdsourced human annotators. 5 F1@K: EXTENDING F1 WITH RECALL FROM HUMAN-PREFERRED LENGTH Finally, we combine factual precision and recall using standard F1, as shown as Equation (1). This proposed metric is bounded between [0, 1], where a higher number indicates a more-factual response. For a response to receive a full score of 1, it must not contain not-supported facts and must have at least K supported facts. F1@K also follows expectations for many clear hypothetical cases, as shown in Appendix D.2, and we contend that it allows for standardized and quantifiable comparisons between language models and between methods of obtaining responses. Equation (1): F1@K measures the long-form factuality of a model response y given the number of supported facts S(y) and the number of not-supported facts N(y) that are in y. The hyperparameter K indicates the number of supported facts required for a response to achieve full recall. We discuss how K should be selected in Appendix D.3. This paper is available on arxiv under CC by 4.0 Deed license. This paper is available on arxiv under CC by 4.0 Deed license. available on arxiv available on arxiv [6] We show performance of step 1 of SAFE (splitting a response into individual facts) in Appendix C.4. Authors:
(1) Jerry Wei, Google DeepMind and a Lead contributors;
(2) Chengrun Yang, Google DeepMind and a Lead contributors;
(3) Xinying Song, Google DeepMind and a Lead contributors;
(4) Yifeng Lu, Google DeepMind and a Lead contributors;
(5) Nathan Hu, Google DeepMind and Stanford University;
(6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign;
(7) Dustin Tran, Google DeepMind;
(8) Daiyi Peng, Google DeepMind;
(9) Ruibo Liu, Google DeepMind;
(10) Da Huang, Google DeepMind;
(11) Cosmo Du, Google DeepMind;
(12) Quoc V. Le, Google DeepMind. Authors: Authors: (1) Jerry Wei, Google DeepMind and a Lead contributors; (2) Chengrun Yang, Google DeepMind and a Lead contributors; (3) Xinying Song, Google DeepMind and a Lead contributors; (4) Yifeng Lu, Google DeepMind and a Lead contributors; (5) Nathan Hu, Google DeepMind and Stanford University; (6) Jie Huang, Google DeepMind and University of Illinois at Urbana-Champaign; (7) Dustin Tran, Google DeepMind; (8) Daiyi Peng, Google DeepMind; (9) Ruibo Liu, Google DeepMind; (10) Da Huang, Google DeepMind; (11) Cosmo Du, Google DeepMind; (12) Quoc V. Le, Google DeepMind.

Stories, News & Mentions about AUTORABIT

Categories

Trending Topics

Stories, News & Mentions about AUTORABIT

Autorabit’s Stories on HackerNoon

Autorabit's News & Mentions

Categories

Trending Topics