Downloads · 30 days
4
9% of all-time downloads
nano-30b-grpo-step10
AbDhumal/nano-30b-grpo-step10
nano-30b-grpo-step10 is a machine learning model from AbDhumal. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for peft.
LoRA adapter from GRPO training on /rfe.speedrun (SFT v4 → GRPO).
All-time downloads
47
Public
Repo size
14.1 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors14.1 GB · 100%
At a glance
- Library
- peft
- Access
- Public
- Created
- Jul 16, 2026
- Updated
- Jul 16, 2026
- SHA
- d1e30e02
Base models
Model card
From the Hugging Face model README
Nemotron Nano 30B GRPO LoRA — step 10
LoRA adapter from GRPO training on /rfe.speedrun (SFT v4 → GRPO).
- Base:
nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 - LoRA: dim=128, alpha=512, exclude
*out_proj* - Checkpoint: GRPO global step 10
- Training: NeMo RL 0.6.0, 8×A100, recipe-prod
Eval note (Jul 16 Harbor 5-way)
- Step 10: best pipeline/tool reliability (
wrote_any_file/pipeline_flow= 100%) - Step 20: best held-out content quality (OQ 2.05, IA 1.35); preferred demo checkpoint
Load with PEFT / vLLM --lora-modules pointing at this adapter directory.
- Library
- peft
- Created
- Jul 16, 2026
- Updated
- Jul 16, 2026