Replace Quantization Modifier

neuralmagic / sparseml

Libraries for applying sparsification recipes to neural networks with a few lines of code, enabling faster and smaller models

Apache License 2.0

2.01k stars 140 forks source link

Closed Satrat closed 1 month ago

Satrat commented 1 month ago

Old QuantizationModifier -> LegacyQuantizationModifier New vLLMQuantizationModifier -> QuantizationModifier

Leaving the old modifier in tact for now for sake of perplexity comparison tests, in a future release we can remove it completely