Downloads · 30 days
0
nemotron30b-1830-grpo-adapters
GoodStartLabs/nemotron30b-1830-grpo-adapters
nemotron30b-1830-grpo-adapters is a machine learning model from GoodStartLabs. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Private backup of the two trained checkpoints from the 1830-finance transfer program. - grpo140/ = run-2 (single-turn GRPO on calibrated 1830 tasks), step 140. W&B q4gbj7bh. - step100/ = run-3 (agentic terminal GRPO,…
Access
Public
Updated Jul 7, 2026
Repo size
3.1 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors3.1 GB · 100%
At a glance
- Access
- Public
- Created
- Jul 7, 2026
- Updated
- Jul 7, 2026
- SHA
- 1e51d628
Model card
From the Hugging Face model README
1830-GRPO LoRA adapters (Nemotron-3-Nano-30B-A3B)
Private backup of the two trained checkpoints from the 1830->finance transfer program.
grpo140/= run-2 (single-turn GRPO on calibrated 1830 tasks), step 140. W&B q4gbj7bh.step100/= run-3 (agentic terminal GRPO, run3-agentic-002), step 100. Base model: nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16. LoRA rank 32 (Tinker export). Merge with Tinker build_hf_model (vLLM LoRA serving FAILS on NemotronH Mamba projections - full-weight merge required). Benchmark provenance: finance_tests/18xx + 18xx-harbor-env worktree, H200_BENCH_PREREG.md.
- Created
- Jul 7, 2026
- Updated
- Jul 7, 2026