LeTI: Learning to Generate from Textual Interactions

• URL: http://arxiv.org/abs/2305.10314v2
• Date: Tue, 19 Mar 2024 11:53:15 GMT
• Title: LeTI: Learning to Generate from Textual Interactions
• Authors: Xingyao Wang, Hao Peng, Reyhaneh Jabbarvand, Heng Ji,
• Abstract summary: We explore LMs' potential to learn from textual interactions (LETI) that not only check their correctness with binary labels but also pinpoint and explain errors in their outputs through textual feedback. Our focus is the code generation task, where the model produces code based on natural language instructions. LETI iteratively fine-tunes the model, using the objective LM, on a concatenation of natural language instructions, LM-generated programs, and textual feedback.
• Score: 60.425769582343506
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Fine-tuning pre-trained language models (LMs) is essential for enhancing their capabilities. Existing techniques commonly fine-tune on input-output pairs (e.g., instruction tuning) or with numerical rewards that gauge the output quality (e.g., RLHF). We explore LMs' potential to learn from textual interactions (LETI) that not only check their correctness with binary labels but also pinpoint and explain errors in their outputs through textual feedback. Our focus is the code generation task, where the model produces code based on natural language instructions. This setting invites a natural and scalable way to acquire textual feedback: the error messages and stack traces from code execution using a Python interpreter. LETI iteratively fine-tunes the model, using the LM objective, on a concatenation of natural language instructions, LM-generated programs, and textual feedback. Prepended to this fine-tuning text, a binary reward token is used to differentiate correct and buggy solutions. LETI requires no ground-truth outputs for training and even outperforms a fine-tuned baseline that does. LETI not only improves the performance of LMs on a code generation dataset MBPP, but also generalizes to other datasets. Trained on MBPP, it achieves comparable or better performance than the base LMs on unseen problems in HumanEval. Furthermore, compared to binary feedback, we observe that textual feedback leads to improved generation quality and sample efficiency, achieving the same performance with fewer than half of the gradient steps. LETI is equally applicable in natural language tasks when they can be formulated as code generation, which we empirically verified on event argument extraction.

Related papers

• Exploring and Unleashing the Power of Large Language Models in Automated Code Translation [40.25727029618665]
  This paper investigates diverse LLMs and learning-based transpilers for automated code translation tasks. UniTrans is a Unified code Translation framework, applicable to various LLMs. Three recent LLMs of diverse sizes are tested with UniTrans, and all achieve substantial improvements.
  arXiv  Detail & Related papers  (2024-04-23T00:49:46Z)
• Grounding Data Science Code Generation with Input-Output Specifications [32.07033683677839]
  Large language models (LLMs) have recently demonstrated a remarkable ability to generate code from natural language prompts. LLMs can have difficulty aligning their outputs with both the NL prompt and the I/O specification. We propose GIFT4Code, a novel approach for the instruction fine-tuning of LLMs with respect to I/O specifications.
  arXiv  Detail & Related papers  (2024-02-12T21:32:49Z)
• LLMRefine: Pinpointing and Refining Large Language Models via Fine-Grained Actionable Feedback [65.84061725174269]
  Recent large language models (LLM) are leveraging human feedback to improve their generation quality. We propose LLMRefine, an inference time optimization method to refine LLM's output. We conduct experiments on three text generation tasks, including machine translation, long-form question answering (QA), and topical summarization. LLMRefine consistently outperforms all baseline approaches, achieving improvements up to 1.7 MetricX points on translation tasks, 8.1 ROUGE-L on ASQA, 2.2 ROUGE-L on topical summarization.
  arXiv  Detail & Related papers  (2023-11-15T19:52:11Z)
• Preference-grounded Token-level Guidance for Language Model Fine-tuning [105.88789610320426]
  Aligning language models with preferences is an important problem in natural language generation. For LM training, based on the amount of supervised data, we present two *minimalist* learning objectives that utilize the learned guidance. In experiments, our method performs competitively on two distinct representative LM tasks.
  arXiv  Detail & Related papers  (2023-06-01T07:00:07Z)
• Improving Code Generation by Training with Natural Language Feedback [69.52985513422381]
  We formalize an algorithm for learning from natural language feedback at training time instead, which we call learning from Language Feedback (ILF) ILF requires only a small amount of human-written feedback during training and does not require the same feedback at test time, making it both user-friendly and sample-efficient. We use ILF to improve a Codegen-Mono 6.1B model's pass@1 rate by 38% relative (and 10% absolute) on the Mostly Basic Python Problems (MBPP) benchmark.
  arXiv  Detail & Related papers  (2023-03-28T16:15:31Z)
• LEVER: Learning to Verify Language-to-Code Generation with Execution [64.36459105535]
  We propose LEVER, a simple approach to improve language-to-code generation by learning to verify the generated programs with their execution results. Specifically, we train verifiers to determine whether a program sampled from the LLMs is correct or not based on the natural language input, the program itself and its execution results. LEVER consistently improves over the base code LLMs(4.6% to 10.9% with code-davinci) and achieves new state-of-the-art results on all of them.
  arXiv  Detail & Related papers  (2023-02-16T18:23:22Z)
• Language Models of Code are Few-Shot Commonsense Learners [106.1531522893209]
  Given a natural language input, the goal is to generate a graph such as an event -- or a reasoning-graph. Existing approaches serialize the output graph as a flat list of nodes and edges. We show that when we instead frame structured commonsense reasoning tasks as code generation tasks, pre-trained LMs of code are better structured commonsense reasoners than LMs of natural language.
  arXiv  Detail & Related papers  (2022-10-13T16:09:36Z)
• LIFT: Language-Interfaced Fine-Tuning for Non-Language Machine Learning Tasks [22.274913349275817]
  Fine-tuning pretrained language models (LMs) without making any architectural changes has become a norm for learning various language downstream tasks. We propose Language-Interfaced Fine-Tuning (LIFT) to solve non-language downstream tasks without changing the model architecture or loss function. LIFT does not make any changes to the model architecture or loss function, and it relies on the natural language interface.
  arXiv  Detail & Related papers  (2022-06-14T02:41:41Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.