Block-scaled INT8 quantization with stochastic rounding largely preserves FP32-cache accuracy
1 more from this model
Block-scaled INT8 quantization with stochastic rounding largely preserves FP32-cache accuracy
implementation detail · filed under inference & serving
Block-scaled INT8 quantization with stochastic rounding quantizes the Mamba SSM cache from FP32 precision to 8-bit INT8.1,2 Stochastic rounding and mantissa precision keep the quantized cache accurate, outperforming FP8 E4M3 which degrades accuracy.3,4
Trade-off It largely preserves FP32-cache accuracy and verbosity while compressing the cache, but no explicit cost is stated beyond that.5
Summarised by glm-5.3-flash from 2 sources; every sentence is checked against the quote it cites.
we quantize the Mamba SSM cache from the original FP32 precision to lower precisions— Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
Block-scaled INT8 quantization with stochastic rounding— NVIDIA Nemotron 3 Ultra Technical Report
the mantissa precision and stochastic rounding are key ingredients for preserving accuracy after Mamba cache quantization— NVIDIA Nemotron 3 Ultra Technical Report
FP8 E4M3 quantization degrades accuracy, likely because it provides lower effective precision for these cache values than INT8— Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
Block-scaled INT8 quantization with stochastic rounding largely preserves FP32-cache accuracy and verbosity— NVIDIA Nemotron 3 Ultra Technical Report
Also called Block-scaled INT8 quantization.
2 quotes from 2 sources, one card per model, each opening with its strongest.
Block-scaled INT8 quantization with stochastic rounding largely preserves FP32-cache accuracy
Block-scaled INT8 quantization with stochastic rounding largely preserves FP32-cache accuracy