Merge pull request #34 from winglian/v2-fixes

fixes for most recent update

matmul_utils_4bit.py

@@ -1,6 +1,6 @@
 import torch
 import numpy as np
-import quant_cuda
+from gptq_llama import quant_cuda
 
 
 # Global Buffer

requirements.txt

@@ -3,6 +3,7 @@ accelerate
 bitsandbytes
 datasets
 sentencepiece
+safetensors
 git+https://github.com/huggingface/transformers.git
 git+https://github.com/sterlind/GPTQ-for-LLaMa.git@lora_4bit
 git+https://github.com/sterlind/peft.git