Downloads · 30 days
7
10% of all-time downloads
nemotron-30b-crypt-rl-reward031-step5-20260609
matthewagi/nemotron-30b-crypt-rl-reward031-step5-20260609
nemotron-30b-crypt-rl-reward031-step5-20260609 is a machine learning model from matthewagi. 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.
This repository has been overwritten from the prior reward-0.31 step-5 adapter and now contains the step 15 LoRA broadcast from the Prime-RL crypt run.
All-time downloads
69
Public
Repo size
3.6 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors1.8 GB · 100%
At a glance
- Library
- peft
- Access
- Public
- Created
- Jun 9, 2026
- Updated
- Jun 9, 2026
- SHA
- 579ca6d0
Model card
From the Hugging Face model README
Nemotron 30B Crypt RL LoRA - Step 15
This repository has been overwritten from the prior reward-0.31 step-5 adapter and now contains the step 15 LoRA broadcast from the Prime-RL crypt run.
Source run on the PI pod:
/root/prime-rl-es/outputs/nemotron_crypt/rl_static_head_resume_step18plus0_rawfiltered4of4_lag_20steps_20260609_0315
Source broadcast:
run_default/broadcasts/step_15
Run notes:
- Base model path used during RL serving/training:
/root/models/nemotron-30b-base-hf-dense-head-static - LoRA rank/alpha: 32/32
- Target modules include q/k/v/o/in/out/up/down projections and lm_head
- Step 15 logged reward: 0.4593
- Step 15 raw correct: 61/128
- Step 15 true-family correct rollouts: concat 61; add 0; abs_diff 0; mul 0
The previous repo name is retained for continuity with existing references, but the contents are now the step 15 adapter.
- Library
- peft
- Created
- Jun 9, 2026
- Updated
- Jun 9, 2026