PleIAs/CommonLingua-Train
Viewer • Updated • 2.76M • 179 • 13
Aunali
Cossale
AI & ML interests
Text2Image and Text2Text generation.
Recent Activity
liked a dataset 1 day ago
PleIAs/CommonLingua-Train liked a model 2 days ago
SicariusSicariiStuff/Assistant_Pepe_32B liked a model 6 days ago
amaai-lab/apexOrganizations
liked a
dataset 1 day ago
liked a
model 2 days ago
liked a
model 6 days ago
liked a
dataset 15 days ago
reacted to qgallouedec's post with 🔥 18 days ago
Post
1914
TRL v1.2 introduces the SSDTrainer 🚀
Simple Self-Distillation (SSD) from Apple's paper "Embarrassingly Simple Self-Distillation Improves Code Generation" is now available as an experimental trainer in TRL.
The recipe is as minimal as the name suggests: sample completions from the model itself at a training-time temperature, then fine-tune on those raw, unverified samples with plain cross-entropy. No reward model. No verifier. No teacher model. No reinforcement learning. Just prompts and the model.
v1.2 also ships expanded tool-calling support (LLaMA 3.1 / 3.2, DeepSeek-V3), another round of KTO ↔ DPO alignment getting us closer to promoting KTO to stable, a big GRPO simplification for overlong tool results, deprecation of
Full release notes: https://github.com/huggingface/trl/releases/tag/v1.2.0
Simple Self-Distillation (SSD) from Apple's paper "Embarrassingly Simple Self-Distillation Improves Code Generation" is now available as an experimental trainer in TRL.
The recipe is as minimal as the name suggests: sample completions from the model itself at a training-time temperature, then fine-tune on those raw, unverified samples with plain cross-entropy. No reward model. No verifier. No teacher model. No reinforcement learning. Just prompts and the model.
from trl.experimental.ssd import SSDConfig, SSDTrainer
trainer = SSDTrainer(
model="Qwen/Qwen3-4B-Instruct",
args=SSDConfig(temperature=0.6, top_k=20, top_p=0.95),
train_dataset=dataset,
)
trainer.train()v1.2 also ships expanded tool-calling support (LLaMA 3.1 / 3.2, DeepSeek-V3), another round of KTO ↔ DPO alignment getting us closer to promoting KTO to stable, a big GRPO simplification for overlong tool results, deprecation of
use_transformers_paged, and key fixes for VLM response parsing.Full release notes: https://github.com/huggingface/trl/releases/tag/v1.2.0
liked a
model 28 days ago
updated a
dataset 29 days ago
updated a
model 29 days ago
updated a
dataset 30 days ago
published a
model about 1 month ago
liked a
model about 1 month ago
liked a
dataset about 1 month ago
updated 4
datasets about 1 month ago
liked a
model about 1 month ago
updated 3
datasets about 1 month ago