split venv_t5 into tiered 5.3.0/5.5.0 and fix trust_remote_code#4878
Conversation
…emote_code auto-enable for native t5 models
|
Warning You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again! |
for more information, see https://pre-commit.ci
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3400654afc
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
for more information, see https://pre-commit.ci
This reverts commit fb43d46.
…idia namespace guard" This reverts commit fc49ae2.
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
|
Warning You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again! |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 5445fa6e16
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 45e3a02cc5
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
1 participant
Summary
.venv_t5(transformers 5.5.0) into.venv_t5_530(5.3.0) and.venv_t5_550(5.5.0)Gemma4ForConditionalGenerationarchitecture ormodel_type: gemma4) in addition to substring matchingget_transformers_tier()API returns"550","530", or"default"trust_remote_codeauto-enable to Nemotron models only — previously it was set toTruefor allunsloth/*transformers 5.x models (introduced in studio: improve onboarding UX, tooltips, and training defaults #4355, partially fixed for Gemma 4 in fix(studio): don't set trust_remote_code for Gemma 4 training #4795), which bypasses the compiler and disables fused cross-entropy. This fixed Qwen3.5-2b on colab T4. Credits to @mmathew23 for pointing this one out.Resolves #4873
Test plan
./install.sh --localcreates both.venv_t5_530and.venv_t5_550with correct versionstest_transformers_version.py)