Mirrors/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2024-11-29 10:59:32 +01:00
Code Issues Packages Projects Releases Wiki Activity

Don't need that

Browse Source
This commit is contained in:
oobabooga 2023-01-09 16:30:14 -03:00
parent 00a12889e9
commit f0013ac8e9
1 changed files with 2 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
server.py
View File

@ -27,11 +27,6 @@ def load_model(model_name):
print(f"Loading {model_name}...") print(f"Loading {model_name}...")
t0 = time.time() t0 = time.time()
if args.cpu:
dtype = torch.float32
else:
dtype = torch.float16
# Loading the model # Loading the model
if not args.cpu and Path(f"torch-dumps/{model_name}.pt").exists(): if not args.cpu and Path(f"torch-dumps/{model_name}.pt").exists():
print("Loading in .pt format...") print("Loading in .pt format...")
@ -45,9 +40,9 @@ def load_model(model_name):
model = T5ForConditionalGeneration.from_pretrained(Path(f"models/{model_name}")).cuda() model = T5ForConditionalGeneration.from_pretrained(Path(f"models/{model_name}")).cuda()
else: else:
if args.cpu: if args.cpu:
model = AutoModelForCausalLM.from_pretrained(Path(f"models/{model_name}"), low_cpu_mem_usage=True, torch_dtype=dtype) model = AutoModelForCausalLM.from_pretrained(Path(f"models/{model_name}"), low_cpu_mem_usage=True, torch_dtype=torch.float32)
else: else:
model = AutoModelForCausalLM.from_pretrained(Path(f"models/{model_name}"), low_cpu_mem_usage=True, torch_dtype=dtype).cuda() model = AutoModelForCausalLM.from_pretrained(Path(f"models/{model_name}"), low_cpu_mem_usage=True, torch_dtype=torch.float16).cuda()
# Loading the tokenizer # Loading the tokenizer
if model_name.lower().startswith('gpt4chan') and Path(f"models/gpt-j-6B/").exists(): if model_name.lower().startswith('gpt4chan') and Path(f"models/gpt-j-6B/").exists():