Fallback to transformers if hf_bleeding_edge not available

modeling/inference_models/generic_hf_torch/class.py

@@ -7,7 +7,10 @@ import shutil
 from typing import Union
 
 from transformers import GPTNeoForCausalLM, GPT2LMHeadModel, BitsAndBytesConfig
+try:
     from hf_bleeding_edge import AutoModelForCausalLM
+except ImportError:
+    from transformers import AutoModelForCausalLM
 
 from transformers.utils import WEIGHTS_NAME, WEIGHTS_INDEX_NAME, TF2_WEIGHTS_NAME, TF2_WEIGHTS_INDEX_NAME, TF_WEIGHTS_NAME, FLAX_WEIGHTS_NAME, FLAX_WEIGHTS_INDEX_NAME, SAFE_WEIGHTS_NAME, SAFE_WEIGHTS_INDEX_NAME
 

modeling/inference_models/gptq_hf_torch/class.py

@@ -10,8 +10,11 @@ import sys
 from typing import Union
 
 from transformers import GPTNeoForCausalLM, AutoTokenizer, LlamaTokenizer
+try:
     import hf_bleeding_edge
     from hf_bleeding_edge import AutoModelForCausalLM
+except ImportError:
+    from transformers import AutoModelForCausalLM
 
 import utils
 import modeling.lazy_loader as lazy_loader

modeling/inference_models/hf.py

@@ -1,6 +1,9 @@
 import os, sys
 from typing import Optional
+try:
     from hf_bleeding_edge import AutoConfig
+except ImportError:
+    from transformers import AutoConfig
 
 import warnings
 import utils

modeling/inference_models/hf_torch.py

@@ -19,7 +19,10 @@ from transformers import (
     GPT2LMHeadModel,
     LogitsProcessorList,
 )
+try:
     from hf_bleeding_edge import AutoModelForCausalLM
+except ImportError:
+    from transformers import AutoModelForCausalLM
 
 import utils
 import modeling.lazy_loader as lazy_loader