An official website of the United States government
This thesis investigates the computational modeling of belief and related cognitive states as expressed in text and speech. Understanding how speakers or authors convey commitment, certainty, and emotions is crucial for language understanding, yet poses significant challenges for current NLP systems. We present a comprehensive study spanning multiple facets of belief prediction. We begin by re-examining the widely used FactBank corpus, correcting a critical projection error and establishing new state-of-the-art results for author-only belief prediction through multi-task learning and error analysis. We then tackle the more complex task of source-and-target belief prediction, introducing a novel generative framework using Flan-T5. This includes developing a structured database representation for FactBank and proposing a linearized tree generation approach, culminating in the BeLeaf system for visualization and analysis, which achieves state-of-the-art performance on both FactBank and the MDP corpus. With the rise of large language models (LLMs), we investigate their zero-shot capabilities for the source-and-target belief task. We propose Unified and Hybrid prompting frameworks, finding that while current LLMs struggle, particularly with nested beliefs, our Hybrid approach paired with reasoning-focused LLMs achieves new state-of-the-art results on FactBank. Finally, we explore the role of multimodality among multiple cognitive states. We present the first study on multimodal belief prediction using the CB-Prosody corpus, demonstrating that integrating audio features via fine-tuned Whisper models significantly improves performance over text-only BERT models. We further introduce Synthetic Audio Data (SAD), showing that even synthetic audio generated by TTS systems provides orthogonal, beneficial signals for various cognitive state tasks (belief, emotion, sentiment). We conclude by presenting OmniVox, the first systematic evaluation of omni-LLMs for zero-shot emotion recognition directly from audio, demonstrating their competitiveness with fine-tuned models and analyzing their acoustic reasoning capabilities.
more »
« less
Simul-LLM: A Framework for Exploring High-Quality Simultaneous Translation with Large Language Models
Large language models (LLMs) with billions of parameters and pretrained on massive amounts of data are now capable of near or better than state-of-the-art performance in a variety of downstream natural language processing tasks. Neural machine translation (NMT) is one such task that LLMs have been applied to with great success. However, little research has focused on applying LLMs to the more difficult subset of NMT called simultaneous translation (SimulMT), where translation begins before the entire source context is available to the model. In this paper, we address key challenges facing LLMs fine-tuned for SimulMT, validate classical SimulMT concepts and practices in the context of LLMs, explore adapting LLMs that are fine-tuned for NMT to the task of SimulMT, and introduce Simul-LLM, the first open-source fine-tuning and evaluation pipeline development framework for LLMs focused on SimulMT.
more »
« less
- Award ID(s):
- 2223483
- PAR ID:
- 10539868
- Publisher / Repository:
- Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (ACL)
- Date Published:
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Toxic content detection is crucial for online services to remove inappropriate content that violates community standards. To automate the detection process, prior works have proposed varieties of machine learning (ML) approaches to train Language Models (LMs) for toxic content detection. However, both their accuracy and transferability across datasets are limited. Recently, Large Language Models (LLMs) have shown promise in toxic content detection due to their superior zero-shot and few-shot in-context learning ability as well as broad transferability on ML tasks.However, efficiently designing prompts for LLMs remains challenging. Moreover, the high run-time cost of LLMs may hinder their deployments in production. To address these challenges, in this work, we propose BD-LLM, a novel and efficient approach to bootstrapping and distilling LLMs for toxic content detection. Specifically, we design a novel prompting method named Decision-Tree-of-Thought (DToT) to bootstrap LLMs' detection performance and extract high-quality rationales. DToT can automatically select more fine-grained context to re-prompt LLMs when their responses lack confidence. Additionally, we use the rationales extracted via DToT to fine-tune student LMs. Our experimental results on various datasets demonstrate that DToT can improve the accuracy of LLMs by up to 4.6%. Furthermore, student LMs fine-tuned with rationales extracted via DToT outperform baselines on all datasets with up to 16.9% accuracy improvement, while being more than 60x smaller than conventional LLMs. Finally, we observe that student LMs fine-tuned with rationales exhibit better cross-dataset transferability.more » « less
-
Large language models (LLMs) have demonstrated revolutionary capabilities in understanding complex contexts and performing a wide range of tasks. However, LLMs can also answer questions that are unethical or harmful, raising concerns about their applications. To regulate LLMs' responses to such questions, a training strategy called alignment can help. Yet, alignment can be unexpectedly compromised when fine-tuning an LLM for downstream tasks. This paper focuses on recovering the alignment lost during fine-tuning. We observe that there are two distinct directions inherent in an aligned LLM: the aligned direction and the harmful direction. An LLM is inclined to answer questions in the aligned direction while refusing queries in the harmful direction. Therefore, we propose to recover the harmful direction of the fine-tuned model that has been compromised. Specifically, we restore a small subset of the fine-tuned model's weight parameters from the original aligned model using gradient descent. We also introduce a rollback mechanism to avoid aggressive recovery and maintain downstream task performance. Our evaluation on 125 fine-tuned LLMs demonstrates that our method can reduce their harmful rate (percentage of answering harmful questions) from 33.25% to 1.74%, without sacrificing task performance much. In contrast, the existing methods either only reduce the harmful rate to a limited extent or significantly impact the normal functionality. Our code is available at https://github.com/kangyangWHU/LLMAlignmentmore » « less
-
Over the past few years, Large Language Models of Code (Code LLMs) have started to have a significant impact on programming practice. Code LLMs are also emerging as building blocks for research in programming languages and software engineering. However, the quality of code produced by a Code LLM varies significantly by programming language. Code LLMs produce impressive results on high-resource programming languages that are well represented in their training data (e.g., Java, Python, or JavaScript), but struggle with low-resource languages that have limited training data available (e.g., OCaml, Racket, and several others). This paper presents an effective approach for boosting the performance of Code LLMs on low-resource languages using semi-synthetic data. Our approach, called MultiPL-T, generates high-quality datasets for low-resource languages, which can then be used to fine-tune any pretrained Code LLM. MultiPL-T translates training data from high-resource languages into training data for low-resource languages in the following way. 1) We use a Code LLM to synthesize unit tests for commented code from a high-resource source language, filtering out faulty tests and code with low test coverage. 2) We use a Code LLM to translate the code from the high-resource source language to a target low-resource language. This gives us a corpus of candidate training data in the target language, but many of these translations are wrong. 3) We use a lightweight compiler to compile the test cases generated in (1) from the source language to the target language, which allows us to filter our obviously wrong translations. The result is a training corpus in the target low-resource language where all items have been validated with test cases. We apply this approach to generate tens of thousands of new, validated training items for five low-resource languages: Julia, Lua, OCaml, R, and Racket, using Python as the source high-resource language. Furthermore, we use an open Code LLM (StarCoderBase) with open training data (The Stack), which allows us to decontaminate benchmarks, train models without violating licenses, and run experiments that could not otherwise be done. Using datasets generated with MultiPL-T, we present fine-tuned versions of StarCoderBase and Code Llama for Julia, Lua, OCaml, R, and Racket that outperform other fine-tunes of these base models on the natural language to code task. We also present Racket fine-tunes for two very recent models, DeepSeek Coder and StarCoder2, to show that MultiPL-T continues to outperform other fine-tuning approaches for low-resource languages. The MultiPL-T approach is easy to apply to new languages, and is significantly more efficient and effective than alternatives such as training longer.more » « less
-
Multiple choice questions are traditionally expensive to produce. Recent advances in large language models (LLMs) have led to fine-tuned LLMs that generate questions competitive with human-authored questions. However, the relative capabilities of ChatGPT-family models have not yet been established for this task. We present a carefully-controlled human evaluation of three conditions: a fine-tuned, augmented version of Macaw, instruction-tuned Bing Chat with zero-shot prompting, and humanauthored questions from a college science textbook. Our results indicate that on six of seven measures tested, both LLM’s performance was not significantly different from human performance. Analysis of LLM errors further suggests that Macaw and Bing Chat have different failure modes for this task: Macaw tends to repeat answer options whereas Bing Chat tends to not include the specified answer in the answer options. For Macaw, removing error items from analysis results in performance on par with humans for all metrics; for Bing Chat, removing error items improves performance but does not reach human-level performance.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Proceeding:
- The DOI is not currently available.
