Skip to content

Fix stopping strings for llama-3 and phi #6043

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
oobabooga merged 4 commits into from
May 22, 2024
Merged

Fix stopping strings for llama-3 and phi #6043

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
0730c19
Fix stopping strings for llama-3 and phi
oobabooga May 22, 2024
ee67286
Minor fixes
oobabooga May 22, 2024
ce89dbc
Account for llama.cpp loader
oobabooga May 22, 2024
bfb9d7f
Simplify
oobabooga May 22, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
75 changes: 38 additions & 37 deletions modules/chat.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,34 +45,35 @@ def str_presenter(dumper, data):
yaml.representer.SafeRepresenter.add_representer(str, str_presenter)


def get_generation_prompt(renderer, impersonate=False, strip_trailing_spaces=True):
def extract_message_prefix_suffix(renderer, strip_trailing_spaces=True):
'''
Given a Jinja template, reverse-engineers the prefix and the suffix for
an assistant message (if impersonate=False) or an user message
(if impersonate=True)
Given a Jinja template, extracts the prefix and suffix for
an assistant message and a user message. It assumes that they
share the same suffix.
'''

if impersonate:
messages = [
{"role": "user", "content": "<<|user-message-1|>>"},
{"role": "user", "content": "<<|user-message-2|>>"},
]
else:
messages = [
{"role": "assistant", "content": "<<|user-message-1|>>"},
{"role": "assistant", "content": "<<|user-message-2|>>"},
]
messages = [
{"role": "user", "content": "<<|user-message-1|>>"},
{"role": "assistant", "content": "<<|assistant-message-1|>>"},
{"role": "user", "content": "<<|user-message-2|>>"},
{"role": "assistant", "content": "<<|assistant-message-2|>>"},
]

prompt = renderer(messages=messages)
unwanted_suffix = renderer(messages=[])

suffix = prompt.split('<<|assistant-message-2|>>')[1]
if unwanted_suffix != '':
suffix = suffix[:-len(unwanted_suffix)]

suffix_plus_prefix = prompt.split("<<|user-message-1|>>")[1].split("<<|user-message-2|>>")[0]
suffix = prompt.split("<<|user-message-2|>>")[1]
prefix = suffix_plus_prefix[len(suffix):]
prefix_user = prompt.split('<<|assistant-message-1|>>')[1].split('<<|user-message-2|>>')[0][len(suffix):]
prefix_assistant = prompt.split('<<|user-message-1|>>')[1].split('<<|assistant-message-1|>>')[0][len(suffix):]

if strip_trailing_spaces:
prefix = prefix.rstrip(' ')
prefix_user = prefix_user.rstrip(' ')
prefix_assistant = prefix_assistant.rstrip(' ')

return prefix, suffix
return prefix_user, prefix_assistant, suffix


def generate_chat_prompt(user_input, state, **kwargs):
Expand Down Expand Up @@ -125,7 +126,12 @@ def generate_chat_prompt(user_input, state, **kwargs):
messages.append({"role": "user", "content": user_input})

def remove_extra_bos(prompt):
for bos_token in ['<s>', '<|startoftext|>', '<BOS_TOKEN>', '<|endoftext|>']:
if hasattr(shared.tokenizer, 'bos_token_id'):
bos_tokens = [shared.tokenizer.decode(shared.tokenizer.bos_token_id)]
else:
bos_tokens = ['<s>', '<|startoftext|>', '<BOS_TOKEN>']

for bos_token in bos_tokens:
while prompt.startswith(bos_token):
prompt = prompt[len(bos_token):]

Expand All @@ -137,6 +143,9 @@ def make_prompt(messages):
else:
prompt = renderer(messages=messages)

prefix_user, prefix_assistant, suffix = extract_message_prefix_suffix(renderer, strip_trailing_spaces=not _continue)
prefix = prefix_user if impersonate else prefix_assistant

if state['mode'] == 'chat-instruct':
outer_messages = []
if state['custom_system_message'].strip() != '':
Expand All @@ -148,29 +157,25 @@ def make_prompt(messages):
command = command.replace('<|prompt|>', prompt)
command = replace_character_names(command, state['name1'], state['name2'])


if _continue:
prefix = get_generation_prompt(renderer, impersonate=impersonate, strip_trailing_spaces=False)[0]
prefix += messages[-1]["content"]
else:
prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
if not impersonate:
prefix = apply_extensions('bot_prefix', prefix, state)
elif not impersonate:
prefix = apply_extensions('bot_prefix', prefix, state)

outer_messages.append({"role": "user", "content": command})
outer_messages.append({"role": "assistant", "content": prefix})

prompt = instruction_template.render(messages=outer_messages)
suffix = get_generation_prompt(instruct_renderer, impersonate=False)[1]
if len(suffix) > 0:
prompt = prompt[:-len(suffix)]

else:

if _continue:
suffix = get_generation_prompt(renderer, impersonate=impersonate)[1]
if len(suffix) > 0:
prompt = prompt[:-len(suffix)]
else:
prefix = get_generation_prompt(renderer, impersonate=impersonate)[0]
if state['mode'] == 'chat' and not impersonate:
prefix = apply_extensions('bot_prefix', prefix, state)

Expand Down Expand Up @@ -249,15 +254,11 @@ def get_stopping_strings(state):
renderers.append(renderer)

for renderer in renderers:
prefix_bot, suffix_bot = get_generation_prompt(renderer, impersonate=False)
prefix_user, suffix_user = get_generation_prompt(renderer, impersonate=True)

stopping_strings += [
suffix_user + prefix_bot,
suffix_user + prefix_user,
suffix_bot + prefix_bot,
suffix_bot + prefix_user,
]
prefix_user, prefix_assistant, suffix = extract_message_prefix_suffix(renderer)

for item in [suffix + prefix_assistant, suffix + prefix_user, suffix]:
stopping_strings.append(item)
stopping_strings.append(item.rstrip())

if 'stopping_strings' in state and isinstance(state['stopping_strings'], list):
stopping_strings += state.pop('stopping_strings')
Expand Down