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 ChatSummaryMemoryBuffer._summarize_oldest_chat_history #17845

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

fix ChatSummaryMemoryBuffer._summarize_oldest_chat_history #17845

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7a2d86d
fix _summarize_oldest_chat_history
pablonm3 Feb 17, 2025
50fd68d
Merge branch 'run-llama:main' into patch-ChatSummaryMemoryBuffer
pablonm3 Feb 18, 2025
3c8a9e2
remove duped system instruction
pablonm3 Feb 18, 2025
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
11 changes: 8 additions & 3 deletions llama-index-core/llama_index/core/memory/chat_summary_memory_buffer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -269,14 +269,19 @@ def _summarize_oldest_chat_history(
):
return chat_history_to_be_summarized[0]

summarize_prompt = ChatMessage(
summarize_prompt = [ChatMessage(
role=MessageRole.SYSTEM,
content=self.summarize_prompt,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hmm, so self.summarize_prompt is already included when you run self._get_prompt_to_summarize(..) (check the end of that function)

Maybe the only change we needed here changing the summarize prompt to the user role?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the response @logan-markewich
what do you think about my last commit to this branch where I remove the prompt += self.summarize_prompt or "" from _get_prompt_to_summarize ? this way we send the instruction as system role and the context as user role which IMO is better than sending both as user role. also IMO it's better to put the instruction before the context rather than after. What do you think?

)]
summarize_prompt.append(ChatMessage(
role=MessageRole.USER,
content=self._get_prompt_to_summarize(chat_history_to_be_summarized),
)
))
# TODO: Maybe it is better to pass a list of history to llm
r = self.llm.chat(summarize_prompt)
return ChatMessage(
role=MessageRole.SYSTEM,
content=self.llm.chat([summarize_prompt]).message.content,
content=r.message.content,
)

def _get_prompt_to_summarize(
Expand Down