Skip to content
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

fix the tokenize process and prompt template of chatglm3 #2883

Merged
merged 2 commits into from
Jan 7, 2024
Merged

fix the tokenize process and prompt template of chatglm3 #2883

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
378d8b6
fix the tokenize process of chatglm3
Jan 4, 2024
e53ba80
format
Jan 5, 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
4 changes: 2 additions & 2 deletions fastchat/conversation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -172,7 +172,7 @@ def get_prompt(self) -> str:
ret += system_prompt
for role, message in self.messages:
if message:
ret += role + "\n" + " " + message
ret += role + "\n" + message
else:
ret += role
return ret
Expand Down Expand Up @@ -487,7 +487,7 @@ def get_conv_template(name: str) -> Conversation:
register_conv_template(
Conversation(
name="chatglm3",
system_template="<|system|>\n {system_message}",
system_template="<|system|>\n{system_message}",
roles=("<|user|>", "<|assistant|>"),
sep_style=SeparatorStyle.CHATGLM3,
stop_token_ids=[
Expand Down
37 changes: 36 additions & 1 deletion fastchat/model/model_chatglm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,6 +37,31 @@ def process_response(response):
return response


def recover_message_list(prompt):
role_token_pattern = "|".join(
[re.escape(r) for r in ["<|system|>", "<|user|>", "<|assistant|>"]]
)
role = None
last_end_idx = -1
message_list = []
for match in re.finditer(role_token_pattern, prompt):
if role:
messge = {}
if role == "<|system|>":
messge["role"] = "system"
elif role == "<|user|>":
messge["role"] = "user"
else:
messge["role"] = "assistant"
messge["content"] = prompt[last_end_idx + 1 : match.start()]
message_list.append(messge)

role = prompt[match.start() : match.end()]
last_end_idx = match.end()

return message_list


@torch.inference_mode()
def generate_stream_chatglm(
model,
Expand All @@ -54,7 +79,17 @@ def generate_stream_chatglm(
max_new_tokens = int(params.get("max_new_tokens", 256))
echo = params.get("echo", True)

inputs = tokenizer([prompt], return_tensors="pt").to(model.device)
model_type = str(type(model)).lower()
if "peft" in model_type:
model_type = str(type(model.base_model.model)).lower()

if "chatglm3" in model_type:
message_list = recover_message_list(prompt)
inputs = tokenizer.build_chat_input(
query=message_list[-1]["content"], history=message_list[:-1], role="user"
).to(model.device)
else:
inputs = tokenizer([prompt], return_tensors="pt").to(model.device)
input_echo_len = len(inputs["input_ids"][0])

gen_kwargs = {
Expand Down