Downloads · 30 days
62
3% of all-time downloads
Cosmos-Reason2-8B-nvfp4
Firworks/Cosmos-Reason2-8B-nvfp4
Cosmos-Reason2-8B-nvfp4 is a machine learning model from Firworks. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Format: NVFP4 — weights & activations quantized to FP4 with dual scaling. Base model: nvidia/Cosmos-Reason2-8B How it was made: One-shot calibration with LLM Compressor (NVFP4 recipe), long-seq calibration with Rombo-…
All-time downloads
2.4K
Public
Parameters
5.7B
14.3 GB on disk
Likes
2
Public
Repo makeup
Click a slice to open those files.
.safetensors7.5 GB · 100%
Parameter types
How the weights are stored.
U83.5B · 61%
Base models
Datasets
Model card
From the Hugging Face model README
Cosmos-Reason2-8B-nvfp4
Format: NVFP4 — weights & activations quantized to FP4 with dual scaling.
Base model: nvidia/Cosmos-Reason2-8B
How it was made: One-shot calibration with LLM Compressor (NVFP4 recipe), long-seq calibration with Rombo-Org/Optimized_Reasoning.
Notes: Keep
lm_headin high precision; calibrate on long, domain-relevant sequences (256 samples at 4096 max length).
Check the original model card for information about this model.
Running the model with VLLM in Docker
sudo docker run --runtime nvidia --gpus all -p 8000:8000 --ipc=host vllm/vllm-openai:nightly --model Firworks/Cosmos-Reason2-8B-nvfp4 --dtype auto --max-model-len 32768
This was tested on an RTX Pro 6000 Blackwell cloud instance.
If there are other models you're interested in seeing quantized to NVFP4 for use on the DGX Spark, or other modern Blackwell (or newer) cards let me know. I'm trying to make more NVFP4 models available to allow more people to try them out.
- Type
- qwen3_vl
- Created
- Jan 6, 2026
- Updated
- Jan 6, 2026