Show, Don't Tell: Aligning Language Models with Demonstrated Feedback

• URL: http://arxiv.org/abs/2406.00888v1
• Date: Sun, 2 Jun 2024 23:13:56 GMT
• Title: Show, Don't Tell: Aligning Language Models with Demonstrated Feedback
• Authors: Omar Shaikh, Michelle Lam, Joey Hejna, Yijia Shao, Michael Bernstein, Diyi Yang,
• Abstract summary: Demonstration ITerated Task Optimization (DITTO) directly aligns language model outputs to a user's demonstrated behaviors. We evaluate DITTO's ability to learn fine-grained style and task alignment across domains such as news articles, emails, and blog posts.
• Score: 54.10302745921713
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Language models are aligned to emulate the collective voice of many, resulting in outputs that align with no one in particular. Steering LLMs away from generic output is possible through supervised finetuning or RLHF, but requires prohibitively large datasets for new ad-hoc tasks. We argue that it is instead possible to align an LLM to a specific setting by leveraging a very small number ($<10$) of demonstrations as feedback. Our method, Demonstration ITerated Task Optimization (DITTO), directly aligns language model outputs to a user's demonstrated behaviors. Derived using ideas from online imitation learning, DITTO cheaply generates online comparison data by treating users' demonstrations as preferred over output from the LLM and its intermediate checkpoints. We evaluate DITTO's ability to learn fine-grained style and task alignment across domains such as news articles, emails, and blog posts. Additionally, we conduct a user study soliciting a range of demonstrations from participants ($N=16$). Across our benchmarks and user study, we find that win-rates for DITTO outperform few-shot prompting, supervised fine-tuning, and other self-play methods by an average of 19% points. By using demonstrations as feedback directly, DITTO offers a novel method for effective customization of LLMs.

Related papers

• SELF-GUIDE: Better Task-Specific Instruction Following via Self-Synthetic Finetuning [70.21358720599821]
  Large language models (LLMs) hold the promise of solving diverse tasks when provided with appropriate natural language prompts. We propose SELF-GUIDE, a multi-stage mechanism in which we synthesize task-specific input-output pairs from the student LLM. We report an absolute improvement of approximately 15% for classification tasks and 18% for generation tasks in the benchmark's metrics.
  arXiv  Detail & Related papers  (2024-07-16T04:41:58Z)
• Are you still on track!? Catching LLM Task Drift with Activations [55.75645403965326]
  Task drift allows attackers to exfiltrate data or influence the LLM's output for other users. We show that a simple linear classifier can detect drift with near-perfect ROC AUC on an out-of-distribution test set. We observe that this approach generalizes surprisingly well to unseen task domains, such as prompt injections, jailbreaks, and malicious instructions.
  arXiv  Detail & Related papers  (2024-06-02T16:53:21Z)
• Speech Translation with Large Language Models: An Industrial Practice [64.5419534101104]
  We introduce LLM-ST, a novel and effective speech translation model constructed upon a pre-trained large language model (LLM) By integrating the large language model (LLM) with a speech encoder and employing multi-task instruction tuning, LLM-ST can produce accurate timestamped transcriptions and translations. Through rigorous experimentation on English and Chinese datasets, we showcase the exceptional performance of LLM-ST.
  arXiv  Detail & Related papers  (2023-12-21T05:32:49Z)
• Take One Step at a Time to Know Incremental Utility of Demonstration: An Analysis on Reranking for Few-Shot In-Context Learning [23.932500424117244]
  In-Context Learning (ICL) is an emergent capability of Large Language Models (LLMs) Previous studies have shown that using LLMs' outputs as labels is effective in training models to select demonstrations. This paper presents an analysis on different utility functions by focusing on LLMs' output probability given ground-truth output.
  arXiv  Detail & Related papers  (2023-11-16T07:03:54Z)
• LLM-augmented Preference Learning from Natural Language [19.700169351688768]
  Large Language Models (LLMs) are equipped to deal with larger context lengths. LLMs can consistently outperform the SotA when the target text is large. Few-shot learning yields better performance than zero-shot learning.
  arXiv  Detail & Related papers  (2023-10-12T17:17:27Z)
• Unlocking the Potential of User Feedback: Leveraging Large Language Model as User Simulator to Enhance Dialogue System [65.93577256431125]
  We propose an alternative approach called User-Guided Response Optimization (UGRO) to combine it with a smaller task-oriented dialogue model. This approach uses LLM as annotation-free user simulator to assess dialogue responses, combining them with smaller fine-tuned end-to-end TOD models. Our approach outperforms previous state-of-the-art (SOTA) results.
  arXiv  Detail & Related papers  (2023-06-16T13:04:56Z)
• PALR: Personalization Aware LLMs for Recommendation [7.407353565043918]
  PALR aims to combine user history behaviors (such as clicks, purchases, ratings, etc.) with large language models (LLMs) to generate user preferred items. Our solution outperforms state-of-the-art models on various sequential recommendation tasks.
  arXiv  Detail & Related papers  (2023-05-12T17:21:33Z)
• Guiding Large Language Models via Directional Stimulus Prompting [114.84930073977672]
  We introduce Directional Stimulus Prompting, a novel framework for guiding black-box large language models (LLMs) toward specific desired outputs. Instead of directly adjusting LLMs, our method employs a small tunable policy model to generate an auxiliary directional stimulus prompt for each input instance.
  arXiv  Detail & Related papers  (2023-02-22T17:44:15Z)
• Self-Generated In-Context Learning: Leveraging Auto-regressive Language Models as a Demonstration Generator [22.532627423361177]
  Self-generated in-context learning (SG-ICL) generates demonstrations for in-context learning from PLM itself. We show SG-ICL significantly outperforms zero-shot learning and is generally worth approximately 0.6 gold training samples.
  arXiv  Detail & Related papers  (2022-06-16T10:52:13Z)

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.