Dancing in Chains: Reconciling Instruction Following and Faithfulness in Language Models

• URL: http://arxiv.org/abs/2407.21417v1
• Date: Wed, 31 Jul 2024 08:05:04 GMT
• Title: Dancing in Chains: Reconciling Instruction Following and Faithfulness in Language Models
• Authors: Zhengxuan Wu, Yuhao Zhang, Peng Qi, Yumo Xu, Rujun Han, Yian Zhang, Jifan Chen, Bonan Min, Zhiheng Huang,
• Abstract summary: We show that modern language models (LMs) need to follow human instructions while being faithful. We propose a simple yet effective method that relies on Rejection Sampling for Continued Self-instruction Tuning (ReSet) We find that less is more, as training ReSet with high-quality, yet substantially smaller data (three-fold less) yields superior results.
• Score: 34.13519934563742
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Modern language models (LMs) need to follow human instructions while being faithful; yet, they often fail to achieve both. Here, we provide concrete evidence of a trade-off between instruction following (i.e., follow open-ended instructions) and faithfulness (i.e., ground responses in given context) when training LMs with these objectives. For instance, fine-tuning LLaMA-7B on instruction following datasets renders it less faithful. Conversely, instruction-tuned Vicuna-7B shows degraded performance at following instructions when further optimized on tasks that require contextual grounding. One common remedy is multi-task learning (MTL) with data mixing, yet it remains far from achieving a synergic outcome. We propose a simple yet effective method that relies on Rejection Sampling for Continued Self-instruction Tuning (ReSet), which significantly outperforms vanilla MTL. Surprisingly, we find that less is more, as training ReSet with high-quality, yet substantially smaller data (three-fold less) yields superior results. Our findings offer a better understanding of objective discrepancies in alignment training of LMs.

Related papers

• RLEF: Grounding Code LLMs in Execution Feedback with Reinforcement Learning [35.446870721902904]
  Large language models (LLMs) deployed as agents solve user-specified tasks over multiple steps while keeping the required manual engagement to a minimum. We propose an end-to-end reinforcement learning method for teaching models to leverage execution feedback in the realm of code synthesis.
  arXiv  Detail & Related papers  (2024-10-02T23:25:17Z)
• Instruction Pre-Training: Language Models are Supervised Multitask Learners [115.95022434390181]
  In this paper, we propose a framework that augments massive raw corpora with instruction-response pairs to pre-train language models (LMs) In our experiments, we synthesize 200M instruction-response pairs covering 40+ task categories to verify the effectiveness of Instruction Pre-Training.
  arXiv  Detail & Related papers  (2024-06-20T16:55:33Z)
• Instruction Tuning With Loss Over Instructions [42.9106826952674]
  Instruction Modelling (IM) trains LMs by applying a loss function to the instruction and prompt part rather than solely to the output part. We show that, in many scenarios, IM can effectively improve the LM performance on both NLP tasks and open-ended generation benchmarks. Remarkably, in the most advantageous case, IM boosts model performance on AlpacaEval 1.0 by over 100%.
  arXiv  Detail & Related papers  (2024-05-23T10:12:03Z)
• Don't Half-listen: Capturing Key-part Information in Continual Instruction Tuning [13.535110749767451]
  We propose a novel continual instruction tuning method based on Key-part Information Gain (KPIG) Our method computes the information gain on masked parts to dynamically replay data and refine the training objective. Experiments demonstrate our method achieves superior performance on both seen and held-out tasks.
  arXiv  Detail & Related papers  (2024-03-15T06:54:20Z)
• Tuna: Instruction Tuning using Feedback from Large Language Models [74.04950416204551]
  We propose finetuning an instruction-tuned large language model using our novel textitprobabilistic ranking and textitcontextual ranking approaches. Probabilistic ranking enables the instruction-tuned model to inherit the relative rankings of high-quality and low-quality responses from the teacher LLM. On the other hand, learning with contextual ranking allows the model to refine its own response distribution using the contextual understanding ability of stronger LLMs.
  arXiv  Detail & Related papers  (2023-10-20T09:55:06Z)
• TRACE: A Comprehensive Benchmark for Continual Learning in Large Language Models [52.734140807634624]
  Aligned large language models (LLMs) demonstrate exceptional capabilities in task-solving, following instructions, and ensuring safety. Existing continual learning benchmarks lack sufficient challenge for leading aligned LLMs. We introduce TRACE, a novel benchmark designed to evaluate continual learning in LLMs.
  arXiv  Detail & Related papers  (2023-10-10T16:38:49Z)
• Instruction Position Matters in Sequence Generation with Large Language Models [67.87516654892343]
  Large language models (LLMs) are capable of performing conditional sequence generation tasks, such as translation or summarization. We propose enhancing the instruction-following capability of LLMs by shifting the position of task instructions after the input sentences.
  arXiv  Detail & Related papers  (2023-08-23T12:36:57Z)
• TIM: Teaching Large Language Models to Translate with Comparison [78.66926087162672]
  We propose a novel framework using examples in comparison to teach LLMs to learn translation. Our approach involves presenting the model with examples of correct and incorrect translations and using a preference loss to guide the model's learning. Our findings offer a new perspective on fine-tuning LLMs for translation tasks and provide a promising solution for generating high-quality translations.
  arXiv  Detail & Related papers  (2023-07-10T08:15:40Z)
• "It's a Match!" -- A Benchmark of Task Affinity Scores for Joint Learning [74.14961250042629]
  Multi-Task Learning (MTL) promises attractive, characterizing the conditions of its success is still an open problem in Deep Learning. Estimateing task affinity for joint learning is a key endeavor. Recent work suggests that the training conditions themselves have a significant impact on the outcomes of MTL. Yet, the literature is lacking a benchmark to assess the effectiveness of tasks affinity estimation techniques.
  arXiv  Detail & Related papers  (2023-01-07T15:16:35Z)

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.