rixty/KoboldAI-Client
1
1
Fork 0
mirror of https://github.com/KoboldAI/KoboldAI-Client.git synced 2025-06-05 21:59:24 +02:00
Code Issues Projects Releases Wiki Activity

Fix 4bit pt loading, add traceback output to GPT2 fallback

Browse Source
This commit is contained in:
0cc4m
2023-05-13 20:15:11 +02:00
parent a2d01bb9e4
commit 266c0574f6
2 changed files with 5 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
modeling/inference_models/hf_torch.py
View File

@@ -291,7 +291,7 @@ class HFTorchInferenceModel(HFInferenceModel):
logger.error("Invalid load key! Aborting.")
raise
logger.warning(f"Fell back to GPT2LMHeadModel due to {e}")
logger.warning(f"Fell back to GPT2LMHeadModel due to {traceback.format_exc()}")
try:
return GPT2LMHeadModel.from_pretrained(location, **tf_kwargs)
except Exception as e:
@@ -538,7 +538,8 @@ class HFTorchInferenceModel(HFInferenceModel):
try:
f = z.open(f"archive/data/{storage_key}")
except:
f = z.open(f"{zipfolder}/data/{storage_key}")
ziproot = z.namelist()[0].split("/")[0]
f = z.open(f"{ziproot}/data/{storage_key}")
current_offset = 0
if current_offset != model_dict[key].seek_offset:
f.read(model_dict[key].seek_offset - current_offset)