zhengjl/text-generation-webui
Public Access
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Use state as function param

Browse Source
This commit is contained in:
oobabooga
2023-04-05 17:22:05 -03:00
parent 19b516b11b
commit 613996dd01
3 changed files with 71 additions and 57 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
modules/chat.py
View File

@@ -91,7 +91,7 @@ def extract_message_from_reply(reply, name1, name2, stop_at_newline):
reply = fix_newlines(reply)
return reply, next_character_found
def 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, stop_at_newline, chat_prompt_size, chat_generation_attempts=1, regenerate=False, mode="cai-chat", end_of_turn=""):
def chatbot_wrapper(text, max_new_tokens, generation_params, seed, name1, name2, context, stop_at_newline, chat_prompt_size, chat_generation_attempts=1, regenerate=False, mode="cai-chat", end_of_turn=""):
just_started = True
eos_token = '\n' if stop_at_newline else None
name1_original = name1
@@ -126,7 +126,7 @@ def chatbot_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical
cumulative_reply = ''
for i in range(chat_generation_attempts):
reply = None
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}:"]):
for reply in generate_reply(f"{prompt}{' ' if len(cumulative_reply) > 0 else ''}{cumulative_reply}", max_new_tokens, generation_params, seed, eos_token=eos_token, stopping_strings=[f"\n{name1}:", f"\n{name2}:"]):
reply = cumulative_reply + reply
# Extracting the reply
@@ -155,7 +155,7 @@ def chatbot_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typical
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, stop_at_newline, chat_prompt_size, chat_generation_attempts=1, mode="cai-chat", end_of_turn=""):
def impersonate_wrapper(text, max_new_tokens, generation_params, seed, name1, name2, context, stop_at_newline, chat_prompt_size, chat_generation_attempts=1, mode="cai-chat", end_of_turn=""):
eos_token = '\n' if stop_at_newline else None
if 'pygmalion' in shared.model_name.lower():
@@ -169,7 +169,7 @@ def impersonate_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typ
cumulative_reply = ''
for i in range(chat_generation_attempts):
reply = None
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}:"]):
for reply in generate_reply(f"{prompt}{' ' if len(cumulative_reply) > 0 else ''}{cumulative_reply}", max_new_tokens, generation_params, 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, stop_at_newline)
yield reply
@@ -181,11 +181,11 @@ def impersonate_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typ
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, stop_at_newline, chat_prompt_size, chat_generation_attempts=1, mode="cai-chat", end_of_turn=""):
for history in 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, stop_at_newline, chat_prompt_size, chat_generation_attempts, regenerate=False, mode=mode, end_of_turn=end_of_turn):
def cai_chatbot_wrapper(text, max_new_tokens, generation_params, seed, name1, name2, context, stop_at_newline, chat_prompt_size, chat_generation_attempts=1, mode="cai-chat", end_of_turn=""):
for history in chatbot_wrapper(text, max_new_tokens, generation_params, seed, name1, name2, context, stop_at_newline, chat_prompt_size, chat_generation_attempts, regenerate=False, mode=mode, end_of_turn=end_of_turn):
yield chat_html_wrapper(history, name1, name2, mode)
def regenerate_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, stop_at_newline, chat_prompt_size, chat_generation_attempts=1, mode="cai-chat", end_of_turn=""):
def regenerate_wrapper(text, max_new_tokens, generation_params, seed, name1, name2, context, stop_at_newline, chat_prompt_size, chat_generation_attempts=1, mode="cai-chat", end_of_turn=""):
if (shared.character != 'None' and len(shared.history['visible']) == 1) or len(shared.history['internal']) == 0:
yield chat_html_wrapper(shared.history['visible'], name1, name2, mode)
else:
@@ -193,7 +193,7 @@ def regenerate_wrapper(text, max_new_tokens, do_sample, temperature, top_p, typi
last_internal = shared.history['internal'].pop()
# Yield '*Is typing...*'
yield chat_html_wrapper(shared.history['visible']+[[last_visible[0], shared.processing_message]], name1, name2, mode)
for history in chatbot_wrapper(last_internal[0], 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, stop_at_newline, chat_prompt_size, chat_generation_attempts, regenerate=True, mode=mode, end_of_turn=end_of_turn):
for history in chatbot_wrapper(last_internal[0], max_new_tokens, generation_params, seed, name1, name2, context, stop_at_newline, chat_prompt_size, chat_generation_attempts, regenerate=True, mode=mode, end_of_turn=end_of_turn):
shared.history['visible'][-1] = [last_visible[0], history[-1][1]]
yield chat_html_wrapper(shared.history['visible'], name1, name2, mode)