gpt4all.cpp

gpt4all.cpp copied to clipboard

beingPurple

beingPurple

beingPurple

The instruction prompt and response (prompt_inp and response_inp) are leaking into the output a lot. This code will prevent that from happening.

pixelkaiser

actually check for it's existence before setting it fixes ```%% ./chat main: seed = 1680128377 llama_model_load: loading model from 'gpt4all-lora-quantized.bin' - please wait ... Illegal instruction: 4 ```

keen99

This is probably junk.

bmschmidt