Skip to content

LmStudioChat_v1.12

Pre-release
Pre-release

Choose a tag to compare

@peerlabx peerlabx released this 09 Nov 19:34
· 3 commits to main since this release

Keep in mind the whole point of this application is to be able to run large models on powerful GPUs remotely using any Android phone. This also keeps your communication with AI completely private. The real added value and best feature will be voice prompting. A little buggy but working. Added some debugging messages for networking purposes. Sometimes does not select the right model and then will not go into the chat prompt without multiple clicking. In order to make this smooth it will take a bit more iterations. But this is a really good start. Was not able to get expo working correctly on any android device so most of the testing was done with a full build. Kind of a pain to have a pipeline that requires builds all the time maybe someone will be able to get that to work.

  • Start LM Studio
  • Make sure developer is enabled and running
  • Got to Settings and Enable COGS and Serve on Local Network
  • There is a reachable at: http://address:1234 posted in the top right hand corner
  • In the app set profile to the same reachable address
  • When app connects the first thing it will do is request a list of all the available models
  • Select your model and then select the chat menu
  • There is a 30 second timeout which will be increased so keep the chats short for now
  • Report more problems to peerlabx/issues