Mirrors/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2025-01-11 21:10:40 +01:00
Code Issues Packages Projects Releases Wiki Activity

Fix another bug

Browse Source
This commit is contained in:
oobabooga 2023-01-18 22:08:23 -03:00
parent 6c7f187586
commit 8f27d33034
1 changed files with 3 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
server.py
View File

@ -142,11 +142,12 @@ 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 = "" if args.cpu else ".cuda()" cuda = "" if args.cpu else ".cuda()"
if eos_token is not None:
n = tokenizer.encode(eos_token, return_tensors='pt')[0][-1]
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}")
else: else:
n = tokenizer.encode(eos_token, return_tensors='pt')[0][-1]
output = eval(f"model.generate(input_ids, eos_token_id={n}, {preset}){cuda}") output = eval(f"model.generate(input_ids, eos_token_id={n}, {preset}){cuda}")
reply = tokenizer.decode(output[0], skip_special_tokens=True) reply = tokenizer.decode(output[0], skip_special_tokens=True)
@ -240,8 +241,8 @@ elif args.chat or args.cai_chat:
return question return question
def chatbot_wrapper(text, tokens, inference_settings, selected_model, name1, name2, context, check): def chatbot_wrapper(text, tokens, inference_settings, selected_model, name1, name2, context, check):
history.append(['', ''])
question = generate_chat_prompt(text, tokens, name1, name2, context) question = generate_chat_prompt(text, tokens, name1, name2, context)
history.append(['', ''])
eos_token = '\n' if check else None eos_token = '\n' if check else None
for i in generate_reply(question, tokens, inference_settings, selected_model, eos_token=eos_token): for i in generate_reply(question, tokens, inference_settings, selected_model, eos_token=eos_token):
reply = i[0] reply = i[0]