Nemotron Labs Dory 1.0 4B
A 4-billion-parameter language model released by NVIDIA, offered in multiple quantization formats including BF16, NVFP4, MXFP8 and GGUF, used for inference performance evaluation.
Timeline (last 90 days)
NVIDIA released the BF16, NVFP4, MXFP8 and GGUF versions of Nemotron Labs Dory 1.0 4B on October 16.
In the GSM8K test (1319 questions, 8192-token limit), BF16 scored 1249, FP8 scored 1241, MXFP8 scored 1250, and NVFP4 scored 1238, with the three quantized formats matching BF16 within a ±3 point tolerance.
All Facts (2)
In the GSM8K test (1319 questions, 8192-token limit), BF16 scored 1249, FP8 scored 1241, MXFP8 scored 1250, and NVFP4 scored 1238, with the three quantized formats matching BF16 within a ±3 point tolerance.
50%UnverifiedNVIDIA released the BF16, NVFP4, MXFP8 and GGUF versions of Nemotron Labs Dory 1.0 4B on October 16.
50%