Mirrors/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2024-11-25 17:29:22 +01:00
Code Issues Packages Projects Releases Wiki Activity

Fix chat_generation_attempts

Browse Source
This commit is contained in:
oobabooga 2023-03-24 02:03:30 -03:00 committed by GitHub
parent 04417b658b
commit 4f5c2ce785
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 13 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
modules/chat.py
View File

@ -115,9 +115,10 @@ def chatbot_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical
yield shared.history['visible']+[[visible_text, shared.processing_message]] yield shared.history['visible']+[[visible_text, shared.processing_message]]
# Generate # Generate
reply = '' cumulative_reply = ''
for i in range(chat_generation_attempts): for i in range(chat_generation_attempts):
for reply in generate_reply(f"{prompt}{' ' if len(reply) > 0 else ''}{reply}", max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, encoder_repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, seed, eos_token=eos_token, stopping_strings=[f"\n{name1}:", f"\n{name2}:"]): for reply in generate_reply(f"{prompt}{' ' if len(cumulative_reply) > 0 else ''}{cumulative_reply}", max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, encoder_repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, seed, eos_token=eos_token, stopping_strings=[f"\n{name1}:", f"\n{name2}:"]):
reply = cumulative_reply + reply
# Extracting the reply # Extracting the reply
reply, next_character_found = extract_message_from_reply(reply, name1, name2, check) reply, next_character_found = extract_message_from_reply(reply, name1, name2, check)
@ -142,6 +143,8 @@ def chatbot_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical
if next_character_found: if next_character_found:
break break
cumulative_reply = reply
yield shared.history['visible'] yield shared.history['visible']
def impersonate_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, encoder_repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, seed, name1, name2, context, check, chat_prompt_size, chat_generation_attempts=1): def impersonate_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, encoder_repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, seed, name1, name2, context, check, chat_prompt_size, chat_generation_attempts=1):
@ -152,15 +155,20 @@ def impersonate_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typ
prompt = generate_chat_prompt(text, max_new_tokens, name1, name2, context, chat_prompt_size, impersonate=True) prompt = generate_chat_prompt(text, max_new_tokens, name1, name2, context, chat_prompt_size, impersonate=True)
reply = ''
# Yield *Is typing...* # Yield *Is typing...*
yield shared.processing_message yield shared.processing_message
cumulative_reply = ''
for i in range(chat_generation_attempts): for i in range(chat_generation_attempts):
for reply in generate_reply(prompt+reply, max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, encoder_repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, seed, eos_token=eos_token, stopping_strings=[f"\n{name1}:", f"\n{name2}:"]): for reply in generate_reply(f"{prompt}{' ' if len(cumulative_reply) > 0 else ''}{cumulative_reply}", max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, encoder_repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, seed, eos_token=eos_token, stopping_strings=[f"\n{name1}:", f"\n{name2}:"]):
reply = cumulative_reply + reply
reply, next_character_found = extract_message_from_reply(reply, name1, name2, check) reply, next_character_found = extract_message_from_reply(reply, name1, name2, check)
yield reply yield reply
if next_character_found: if next_character_found:
break break
cumulative_reply = reply
yield reply yield reply
def cai_chatbot_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, encoder_repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, seed, name1, name2, context, check, chat_prompt_size, chat_generation_attempts=1): def cai_chatbot_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical_p, repetition_penalty, encoder_repetition_penalty, top_k, min_length, no_repeat_ngram_size, num_beams, penalty_alpha, length_penalty, early_stopping, seed, name1, name2, context, check, chat_prompt_size, chat_generation_attempts=1):