Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

Appearance settings

MiniCPM-V 2.6 memory leak occurred !!! #1886

Copy link
Copy link
Open
@Liwx1014

Description

@Liwx1014
Issue body actions

Prerequisites

Please answer the following questions for yourself before submitting an issue.

  • I am running the latest code. Development is very rapid so there are no tagged versions as of now.
  • [ ✅] I carefully followed the README.md.
  • [ ✅] I searched using keywords relevant to my issue to make sure that I am creating a new issue that is not already open (or closed).
  • [ ✅] I reviewed the Discussions, and have a new bug or useful enhancement to share.

Expected Behavior

MiniCPM-V 2.6 memory leak ?

Current Behavior

I am testing minicpm-v-2.6 ,here is part of my test code:
37a5d00121d9619b47c64c17b35131b

when i run code ,it adds 10MB per inference,i use memory_profiler find in

embed = self._embed_image_bytes(image_bytes, llama.context_params.n_threads_batch)

a6879c9732cf5cd576859e126b66faf

I thought at first that this variable "embed" wasn't being released,I added the code manual release on line 2856,but the program reported an error.
then I replace "embed" with "self._last_image_embed",Memory leaks continue to occur.

Environment and Context

Please provide detailed information about your computer setup. This is important in case the issue is not reproducible except for under certain specific conditions.then

Example environment info:


Python 3.10.10
ubuntu 18.04
cudatoolkit 12.1
llama-cpp-python 0.2.90

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions

      Morty Proxy This is a proxified and sanitized view of the page, visit original site.