Hugging Face's logo

Collections

Discover the best community collections!

Collections trending this week
OnQ
OnQ: Qwen3-1.7B, 4B and 8B at W2/W3, group size 128. On-policy distillation checkpoints released as BF16 fake-quant inference weights.
EM organisms: LoRA sweep (single vs every layer vs full FT)
#10 LoRA sweep: rank x layer single-layer grid, every-layer LoRA, twins, stacking, and the matching full fine-tunes
OnQ
OnQ: Qwen3-1.7B, 4B and 8B at W2/W3, group size 128. On-policy distillation checkpoints released as BF16 fake-quant inference weights.
EM organisms: LoRA sweep (single vs every layer vs full FT)
#10 LoRA sweep: rank x layer single-layer grid, every-layer LoRA, twins, stacking, and the matching full fine-tunes