Need help training with fine-tuning (pls)

[voltioDS3]

Long story short: my friend and I have fine-tuned with a custom data.json till checkpoint-200 but now we don't know how to run it. We dont really understand how use the model we trained to input prompts and get answers, we just know that after running finetune.py a folder was created with the name "lora-alpaca", inside it has a "checkpoint-200" folder, a "run" folder, an "adapter_config.json" and an "adapter_model.bin". Please we would be really thankful if someone could help us.

[AngainorDev]

The final model is the two adapter_. files.

Use generate.py to run your lora.
Change the lora path from LORA_WEIGHTS = "tloen/alpaca-lora-7b" to LORA_WEIGHTS = "./lora-alpaca" , you local lora.

[OfficerChul]

I think you should also specify the checkpoint path for example LORA_WEIGHTS = "./lora-alpaca/checkpoint-200"