Downloads · 30 days
22
42% of all-time downloads
geodesic-research/nemotron-super-120b-cc-mt-replay-only-1b-200k-sft
nemotron-super-120b-cc-mt-replay-only-1b-200k-sft is a machine learning model from geodesic-research. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as other.
Nemotron 3 Super 120B-A12B non-reasoning (direct-answer chat) SFT from the constitutional-curriculum midtraining study — the SFT stage applied to the cc-mt-replay-only-1b midtrained base (the no-intervention baseline:…
Downloads · 30 days
22
42% of all-time downloads
All-time downloads
53
Public
Parameters
121B
241 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors241 GB · 100%
How the weights are stored.
BF16121B · 100%
From the Hugging Face model README
Nemotron 3 Super 120B-A12B non-reasoning (direct-answer chat) SFT from the constitutional-curriculum midtraining study —
the SFT stage applied to the cc-mt-replay-only-1b midtrained base (the no-intervention baseline: 1.0B tokens of
pretraining replay only, no curriculum data).
no_thinkgeodesic-research/nemotron-instruct-tokenizer, answer-only loss. Final lm loss 0.6251.
Training run: W&B fkmis4gw.The original July 2026 upload of this repo was lost. These weights are the original trained weights, recovered from a Megatron conversion cache made on 2026-07-09 from the original HF export and re-exported to HF. Verified against an independent July copy (the GSM8k LoRA merge of this model): every tensor not touched by that LoRA is bitwise-identical. Config, tokenizer and chat template are the original July export files.
Chat model — apply the bundled chat template. Recommended sampling: temperature 0.6, top_p 0.95
(Nemotron recommendation; higher temperatures occasionally emit a stray token right before EOS).
Loads with native transformers NemotronHForCausalLM; tokenizer PreTrainedTokenizerFast.