qgallouedec/Qwen3-0.6B-SFT-20251113165959
Text Generation • 0.6B • Updated • 291
Quentin Gallouédec PRO
qgallouedec
AI & ML interests
None yet
Recent Activity
posted an update about 23 hours ago
TRL v1.3 ships day-one training support for Qwen 3.6 🚀
The new Qwen 3.6 family (`Qwen/Qwen3.6-27B`, `Qwen/Qwen3.6-35B-A3B`) reuses the Qwen3.5-MoE architecture but ships a slightly different chat template, so we updated the stack end-to-end: new training template with `{% generation %}` markers, tool-call response schema routing, tiny test models for the VLM matrix.
SFT with assistant-only loss works out of the box:
```python
from trl import SFTConfig, SFTTrainer
trainer = SFTTrainer(
model="Qwen/Qwen3.6-27B",
args=SFTConfig(assistant_only_loss=True),
train_dataset=dataset,
)
trainer.train()
```
So does GRPO tool-calling — just hand `tools=[...]` to `GRPOTrainer`.
v1.3 also brings a new experimental TPO trainer (Triple Preference Optimization), speculative decoding in `trl vllm-serve` (Qwen3 MTP / Eagle3 drafts), 12 more KTO ↔ DPO alignment PRs (KTO promotion to stable is now in reach), three more `{% generation %}` chat templates (Gemma/Gemma 2, Phi-3, GLM-4-MoE), and a chunky SFT entropy bug fix.
Full release notes: https://github.com/huggingface/trl/releases/tag/v1.3.0
updated a dataset 1 day ago
hf-doc-build/doc-build updated a bucket 1 day ago
hf-doc-build/docOrganizations
qgallouedec/tiny-aya-global-SFT
Updated
qgallouedec/tiny-aya-global-tool-calling-SFT
Updated
qgallouedec/my-other-awesome-model
Text Generation • 0.5B • Updated • 12
qgallouedec/my-awesome-model
Text Generation • 0.5B • Updated • 13
qgallouedec/trainer_output
Text Generation • 0.5B • Updated • 10
qgallouedec/test_push_output_4
Text Classification • 87.5k • Updated • 6
qgallouedec/qwen2-0.5b-deepmath-grpo
Updated
qgallouedec/my-finetuned-model
0.8B • Updated • 1
qgallouedec/Qwen3-0.6B-SFT-20251113163732
Updated
qgallouedec/Meta-Llama-3-8B-Instruct-SFT-20251112173255
Updated
qgallouedec/Meta-Llama-3-8B-Instruct-SFT-20251112165832
Updated
qgallouedec/Meta-Llama-3-8B-Instruct-SFT-20251112171926
Updated
qgallouedec/Meta-Llama-3-8B-Instruct-SFT-20251112171823
Updated
qgallouedec/gold-model
Updated
qgallouedec/custom-resnet50d
Feature Extraction • 25.6M • Updated • 4
qgallouedec/Qwen3-1.7B-parsing
Text Generation • 2B • Updated • 5
qgallouedec/Qwen2.5-0.5B-SFT
Text Generation • 0.5B • Updated • 7
qgallouedec/Qwen2-0.5B-Reward
Token Classification • 0.5B • Updated • 6
qgallouedec/Qwen3-0.6B-SFT-20250911031144
Text Generation • 0.6B • Updated • 10
qgallouedec/Qwen3-0.6B-SFT-20250911023224
Text Generation • 0.6B • Updated • 6
qgallouedec/Qwen3-0.6B-Base-SFT-20250911020040
Text Generation • 0.6B • Updated • 4
qgallouedec/Qwen3-0.6B-SFT-20250911021538
Text Generation • 0.6B • Updated • 2
qgallouedec/Qwen3-0.6B-Base-SFT-20250911021314
Text Generation • 0.6B • Updated • 3
qgallouedec/Qwen3-0.6B-Base-SFT-20250911014759
Text Generation • 0.6B • Updated • 2
qgallouedec/Qwen3-0.6B-Base-SFT-20250911011255
Text Generation • 0.6B • Updated • 4
qgallouedec/after
Text Generation • 0.5B • Updated • 7
qgallouedec/before
Text Generation • 0.5B • Updated • 7
qgallouedec/Qwen3-1.7B-SFT-20250910184326
Text Generation • 2B • Updated • 10
qgallouedec/Qwen3-4B-SFT-20250910180651
Text Generation • 4B • Updated • 2