hf-doc-build/doc-build
Updated • 306k • 38
Quentin Gallouédec PRO
qgallouedec
AI & ML interests
None yet
Recent Activity
updated a dataset about 17 hours ago
hf-doc-build/doc-build updated a bucket about 17 hours ago
hf-doc-build/doc updated a model 1 day ago
qgallouedec/Qwen3-4B-Thinking-2507-noisyOrganizations
updated a
dataset about 17 hours ago
updated a
bucket about 17 hours ago
updated a
model 1 day ago
published a
model 1 day ago
updated a
bucket 1 day ago
liked a
dataset 1 day ago
liked a
model 1 day ago
posted an update 3 days ago
Post
9588
Shipped hf-sandbox! 🥡
🧪 Running an eval that executes model-generated C on a few thousand prompts? You probably don't want any of that on your laptop.
Just shipped hf-sandbox, a Modal-style sandbox API on top of Hugging Face Jobs. Spin up an isolated, ephemeral container, run untrusted code, get the result back. No Docker on your laptop, no infra to manage.
Just pip install hf-sandbox.
Early days (v0.1); feedback and issues very welcome:
👉 https://github.com/huggingface/hf-sandbox
🧪 Running an eval that executes model-generated C on a few thousand prompts? You probably don't want any of that on your laptop.
Just shipped hf-sandbox, a Modal-style sandbox API on top of Hugging Face Jobs. Spin up an isolated, ephemeral container, run untrusted code, get the result back. No Docker on your laptop, no infra to manage.
Just pip install hf-sandbox.
Early days (v0.1); feedback and issues very welcome:
👉 https://github.com/huggingface/hf-sandbox
- 1 reply
Upload Qwen2_5_VLForConditionalGeneration
#11 opened 4 days ago
by 
qgallouedec
Upload Qwen2_5_VLForConditionalGeneration
#10 opened 4 days ago
by qgallouedec
Upload Qwen2_5_VLForConditionalGeneration
#9 opened 4 days ago
by qgallouedec
posted an update 4 days ago
Post
174
**TRL v1.4 is out 🚀** Chunked NLL loss for SFT and a first-class **OpenReward** integration.
**Chunked NLL loss for SFT — drops peak VRAM by up to 14×**
Standard SFT materializes a full
Peak GPU memory, AdamW fp32:
- Qwen3-14B, 8×H100 FSDP2, 16k seq: 58.9 GB → 38.9 GB
- Qwen3-4B, 1×H100 80GB, 16k seq: OOM → 63.8 GB
- Qwen3-32B, 8×H100 FSDP2, 8k seq: OOM → 71.2 GB
End-to-end it's consistently as fast or faster than
Works with PEFT and VLMs out of the box.
**Open Reward Standard environment adapter**
The new
v1.4 also brings MFU helpers for dense + MoE models, GRPO support for Liger 0.8.0 (delta clipping + VESPO + KL bias correction), Tülu 3's length-normalized DPO loss, four more training chat templates (Cohere, Cohere2, Gemma 3, Qwen3-2507), and a 5+ GB CUDA memory leak fix in activation offloading.
Full release notes: https://github.com/huggingface/trl/releases/tag/v1.4.0
**Chunked NLL loss for SFT — drops peak VRAM by up to 14×**
Standard SFT materializes a full
[batch × seq × vocab] logits tensor before computing cross-entropy, which dominates peak memory at long context lengths. The new loss_type="chunked_nll" path drops ignored-label tokens before the lm_head matmul and computes cross-entropy in checkpointed chunks of 256.Peak GPU memory, AdamW fp32:
- Qwen3-14B, 8×H100 FSDP2, 16k seq: 58.9 GB → 38.9 GB
- Qwen3-4B, 1×H100 80GB, 16k seq: OOM → 63.8 GB
- Qwen3-32B, 8×H100 FSDP2, 8k seq: OOM → 71.2 GB
End-to-end it's consistently as fast or faster than
nll, and unlocks sequence lengths that don't fit at all under the standard path.SFTConfig(loss_type="chunked_nll")Works with PEFT and VLMs out of the box.
**Open Reward Standard environment adapter**
The new
trl.experimental.openreward adapter plugs any environment speaking the [Open Reward Standard](https://openrewardstandard.io) protocol into any TRL trainer that takes an environment_factory. One string — a catalog name or a URL — wires the dataset, factory, and reward_func slots; tools are bound dynamically from JSON Schema, no per-env wrapper code:from trl import GRPOTrainer
from trl.experimental.openreward import OpenRewardSpec
spec = OpenRewardSpec("Eigent/SETA", num_tasks=64)
trainer = GRPOTrainer(
...,
train_dataset=spec.train_dataset,
environment_factory=spec.environment_factory,
reward_funcs=spec.reward_funcs,
)v1.4 also brings MFU helpers for dense + MoE models, GRPO support for Liger 0.8.0 (delta clipping + VESPO + KL bias correction), Tülu 3's length-normalized DPO loss, four more training chat templates (Cohere, Cohere2, Gemma 3, Qwen3-2507), and a 5+ GB CUDA memory leak fix in activation offloading.
Full release notes: https://github.com/huggingface/trl/releases/tag/v1.4.0
upvoted an article 5 days ago
Article
EMO: Pretraining mixture of experts for emergent modularity
allenai
• • 30 Fix chat_template crash when assistant message omits the `content` key
#35 opened 6 days ago
by qgallouedec
Fix chat_template crash when assistant message omits the `content` key
#3 opened 6 days ago
by qgallouedec
Fix chat_template crash when assistant message omits the `content` key
#122 opened 6 days ago
by qgallouedec
Fix chat_template crash when assistant message omits the `content` key
#51 opened 6 days ago
by qgallouedec
Fix chat_template crash when assistant message omits the `content` key
#23 opened 6 days ago
by qgallouedec
Fix chat_template crash when assistant message omits the `content` key
#48 opened 6 days ago
by qgallouedec