unslothai/unsloth

DPO, ORPO - grad accumulation fix

Closed

#1,178 opened on Oct 24, 2024

View on GitHubContributor guide
 (1 comment) (0 reactions) (0 assignees)Python (5,658 forks)batch import
feature requesthelp wanted

Repository metrics

Stars
 (64,271 stars)
PR merge metrics
 (Avg merge 3d 15h) (525 merged PRs in 30d)

Description

Goal: Propagate gradient accumulation fix to DPO - much harder since it requires a full rewrite of https://github.com/huggingface/trl/blob/main/trl/trainer/dpo_trainer.py

Contributor guide

Research direction
Study the DPO trainer in trl library, review the gradient accumulation fix already applied elsewhere (e.g., ORPO), and rewrite the DPO trainer to incorporate the fix.
Tech stack
pythonpytorch
Domain
machine learning
Issue type
Bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-2 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Fresh
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Mostly clear
Prerequisites
PythonPyTorchDPO
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
30