Kevin Michael Schindler

@Ryosama2 for me it worked. It looks like you installed the torch cuda version in the base env and then switched to RESRGAN env which might still have the wrong...

I have just removed the check for session_id and now it works fine. Could of course break other things but it unblocked me for now :) src\backend\base\langflow\memory.py ``` if not...

I think I fixed it. It has something to do with the psycopg2 dependency. The current version uses sqlalchemy and the prior versions uses psycopg directly. "psycopg2-binary>=2.9.6", "psycopg>=3.1.9", #"sqlalchemy[aiosqlite,postgresql_psycopgbinary]>=2.0.36", Also...

> @KevinMichaelSchindler awesome! @italojohnny If that would be the issue, I wonder if the hotfix for langflow could be released? I am also using sqlite version The previous langflow version...

I tried to add the aiosqlite manually: "aiosqlite>=0.19.0", But there are still unresolved errors: FAILED src/backend/tests/unit/api/v1/test_users.py::test_add_user - RuntimeError: There's a mismatch between the models and the database. Results (312.75s (0:05:12)):...

You are using sqlite in your .env right? (python3 -m langflow run --env-file ./.env --no-open-browser --auto-saving --host localhost) I just tested it for the "make init" locally and the docker-compose.yml...