datasets:
- Rombo-Org/Optimized_Reasoning
base_model:
- HuggingFaceTB/SmolLM3-3B
license: apache-2.0
SmolLM3-3B-nvfp4
Format: NVFP4 — weights & activations quantized to FP4 with dual scaling.
Base model: HuggingFaceTB/SmolLM3-3B
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.
Check the original model card for information about this model.
Running the model with VLLM in Docker
Note: I can't get this one to run. The quantization appeared to complete successfully but it won't load for me in VLLM. Maybe someone has an idea how to get it working?
sudo docker run --runtime nvidia --gpus all -p 8000:8000 --ipc=host vllm/vllm-openai:nightly --model Firworks/SmolLM3-3B-nvfp4 --dtype auto --max-model-len 32768
This was tested on a B200 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.