|
@@ -141,7 +141,7 @@ def generate_reply(question, tokens, inference_settings, selected_model, eos_tok
|
|
|
|
|
|
|
|
input_ids = encode(question, 1)
|
|
input_ids = encode(question, 1)
|
|
|
preset = preset.replace('max_new_tokens=tokens', 'max_new_tokens=1')
|
|
preset = preset.replace('max_new_tokens=tokens', 'max_new_tokens=1')
|
|
|
- cuda = ".cuda()" if args.cpu else ""
|
|
|
|
|
|
|
+ cuda = ".cuda()" if not args.cpu else ""
|
|
|
for i in range(tokens):
|
|
for i in range(tokens):
|
|
|
if eos_token is None:
|
|
if eos_token is None:
|
|
|
output = eval(f"model.generate(input_ids, {preset}){cuda}")
|
|
output = eval(f"model.generate(input_ids, {preset}){cuda}")
|
oobabooga