how to deny the auto created output.png

THUDM / CogVLM

a state-of-the-art-level open visual language model | 多模态预训练模型

Apache License 2.0

5.86k stars 400 forks source link

how to deny the auto created output.png #416

Closed HarryXuancy closed 3 months ago

HarryXuancy commented 5 months ago

System Info / 系統信息

does not matter

Who can help? / 谁可以帮助到您？

No response

Information / 问题信息

[X] The official example scripts / 官方的示例脚本
[ ] My own modified scripts / 我自己修改的脚本和任务

Reproduction / 复现过程

1.run the cli_demo_sat.py

an output.png is created in basic_demo/

Expected behavior / 期待表现

I wonder where does the png generated and want to deny it. Thank you.

zRzRzRzRzRzRzR commented 5 months ago

This model is an image understanding model, not an image generation model, and cannot generate pictures.

HarryXuancy commented 5 months ago

Hello, maybe you can pay attention to my description in "Reproduction". I notice that every time I finished the conversation with model, there will be an output.png in the basic_demo folder, which is the last picture that i showed to the cogagent. I guess it is recorded during the response process but i don't know where the exact code is. By the way, i use the sat version cli, the exact command is "python cli_demo_sat.py --from_pretrained cogagent-vqa --version chat_old --bf16 --stream_chat"