Post by avovk1 in MyRobot [Demo] (MOD SUPPORT Update) comments

MyRobot [Demo] (MOD SUPPORT Update) » Comments

Viewing post in MyRobot [Demo] (MOD SUPPORT Update) comments

Hello!
Currently local hosting LLM via llama.cpp is not working
Requests do come in, being processed, but then they are cancelled client-side.
Is there a way to open console/debug mode to look why exactly?

sgthale28 days ago

yes take a look at the logs in the game's files data/logs/

avovk127 days ago

okay, so there are a lot of errors
1. Player2 is not responding
2. Health request sent to unknown port on localhost, i suppose it tries to determine health of AI inference endpoint (:4315/v1/health)
3. ChatManager gives "Could not fetch token usage! 401"
4. and when game determined that "Local LLM: OFFLINE" - it gave "operation timed out" error (it did send request ignoring previous statement)
fourth error is 10 seconds after request, while my endpoint is still generating (i tested with CPU inference, GPU is unavailable at the moment)

And after that, repeated error of this kind, with few missing animations

avovk127 days ago

Ah.
Also
Half of the file is error about destroying without calling SafeDestroy

sgthale27 days ago

Make sure your ai model is not a reasoning type as it's not supported right now. also double check with the in-game guide. You can ignore the SafeDestroy calls.

avovk123 days ago

Double checked, it still makes calls to unregistered port that does not have a backend to check for heartbeat of inference server, and still times out - do you not use streaming responce?

And there is a lot of errors...

sgthale23 days ago (1 edit)

Okay i'll look at it sometime this week. Thanks for the heads up! Btw most of those errors are benign (mostly debug calls accidentally left in)

itch.io

Viewing post in MyRobot [Demo] (MOD SUPPORT Update) comments