ilotoki_thu/alpaca_lora_4bit
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

fix bug

Browse Source
This commit is contained in:
John Smith
2023-04-26 13:13:54 +08:00
parent 1abdc99675
commit d6791790ed
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
server/server.py
View File

@@ -88,7 +88,7 @@ class ModelServer:
print('Quantized attention applied.') print('Quantized attention applied.')
if self.lora_path is not None: if self.lora_path is not None:
inject_lora_layers(model, self.lora_path, device='cuda', torch_dtype=torch.float16) inject_lora_layers(model, self.lora_path, device='cuda', dtype=torch.float16)
self.model, self.tokenizer = model, tokenizer self.model, self.tokenizer = model, tokenizer
print("Loaded in {:.2f} seconds.".format(time.time() - t0)) print("Loaded in {:.2f} seconds.".format(time.time() - t0))