compressed-tensors
Subscription: Developer
Active
Subscription: Developer
Library for utilizing compressed safetensors of neural network models, supporting quantization and sparsity schemes like GPTQ, AWQ, SmoothQuant, INT8, FP8, SparseGPT, providing efficient storage and management of optimized model data.
pip install compressed_tensors==0.15.0.1
0.15.0.1
Linux