SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.16944 · 2023

Zephyr: Direct Distillation of LM Alignment

Alexander Rush, Nazneen Rajani, Thomas Wolf, Kashif Rasul, Shengyi Huang, Younes Belkada, Leandro Von Werra, Lewis Tunstall, Omar Sanseviero, Nathan Lambert, Clémentine Fourrier, Edward Beeching, and 3 more

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 1 functions out of this paper's own repositories and ran 1 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
copy not recorded — 1 of 1
FunctionStatusWhere it lives
compute_loss Ran this paper's copy was not recorded; identical code first harvested from Savannah120/alignment-handbook-PoFT
pointer only · get_code("d1e156f54d1adeec")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We aim to produce a smaller language model that is aligned to user intent. Previous research has shown that applying distilled supervised fine-tuning (dSFT) on larger models significantly improves task accuracy; however, these models are unaligned, i.e. they do not respond well to natural prompts. To distill this property, we experiment with the use of preference data from AI Feedback (AIF). Starting from a dataset of outputs ranked by a teacher model, we apply distilled direct preference optimization (dDPO) to learn a chat model with significantly improved intent alignment. The approach requires only a few hours of training without any additional sampling during fine-tuning. The final result, ZEPHYR-7B, sets a new state-of-the-art on chat benchmarks for 7B parameter models, and requires no human annotation. In particular, results on MT-Bench show that ZEPHYR-7B surpasses LLAMA2-CHAT-70B, the best open-access RLHFbased model.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2310.16944")
get_code_for_paper("2310.16944")
have("2310.16944")

Connect an agent — have() is free.