fix: Remove chat template setting from non-SFT trainer scripts #4437
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Resolves #4404 - Remove SIMPLE_CHAT_TEMPLATE import from 7 trainer scripts - Remove chat template setting for non-SFT trainers (DPO, CPO, ORPO, PPO, Nash-MD, XPO, Online DPO) - Chat templates only make sense for SFT (instruction tuning), not for preference optimization or reward-based training - Scripts modified: - examples/scripts/online_dpo.py - examples/scripts/orpo.py - examples/scripts/cpo.py - examples/scripts/nash_md.py - examples/scripts/xpo.py - examples/scripts/ppo/ppo.py - examples/scripts/ppo/ppo_tldr.py
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
qgallouedec
added a commit
that referenced
this pull request
Nov 4, 2025
commit 4677cf2 Author: Harras Mansoor <[email protected]> Date: Wed Nov 5 04:06:13 2025 +0500 Removed Sentiment Tuning Examples (#4424) commit 7a9592b Author: Quentin Gallouédec <[email protected]> Date: Tue Nov 4 14:32:04 2025 -0700 🐍 Drop Python 3.9 (#4183) commit 7f15a7f Author: Harras Mansoor <[email protected]> Date: Wed Nov 5 02:06:31 2025 +0500 Removed outdated warning about batch contamination (#4423) commit 8b0a3ce Author: Albert Villanova del Moral <[email protected]> Date: Tue Nov 4 21:37:39 2025 +0100 Update tokenizer apply_chat_template with return_dict=True default (#4448) commit d9f9e2b Author: Pramodith Ballapuram <[email protected]> Date: Tue Nov 4 19:56:58 2025 +0000 Support casting to fp32 when word embeddings are tied to lm_head (#4446) commit 4e138ab Author: Sergio Paniego Blanco <[email protected]> Date: Tue Nov 4 15:15:23 2025 +0100 Upload notebook with T4 selected (#4449) commit 43253b2 Author: Pramodith Ballapuram <[email protected]> Date: Mon Nov 3 21:07:31 2025 +0000 Add On-Policy Distillation from thinking labs to paper index. (#4410) Co-authored-by: Quentin Gallouédec <[email protected]> commit 6f41b18 Author: Behrooz Azarkhalili <[email protected]> Date: Mon Nov 3 10:57:51 2025 -0800 fix: Remove chat template setting from non-SFT trainer scripts (#4437) Co-authored-by: Quentin Gallouédec <[email protected]> Co-authored-by: Quentin Gallouédec <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
Resolves #4404
This PR removes the
SIMPLE_CHAT_TEMPLATEimport and chat template setting from all non-SFT trainer scripts. Setting chat templates only makes sense for SFT (supervised fine-tuning/instruction tuning), not for preference optimization or reward-based training methods.Changes
Removed chat template setting from:
examples/scripts/online_dpo.py- Online DPO (preference optimization)examples/scripts/orpo.py- ORPO (preference optimization)examples/scripts/cpo.py- CPO (preference optimization)examples/scripts/nash_md.py- Nash-MD (multi-objective RL)examples/scripts/xpo.py- XPO (preference optimization)examples/scripts/ppo/ppo.py- PPO (reward-based training)examples/scripts/ppo/ppo_tldr.py- PPO TLDR (reward-based training)For each script:
from trl.trainer.utils import SIMPLE_CHAT_TEMPLATEimporttokenizer.chat_template = SIMPLE_CHAT_TEMPLATERationale
Chat templates are used to format conversational data for instruction tuning (SFT). They don't apply to:
Setting chat templates in these contexts was unnecessary and could cause confusion about the expected data format.