# Alpaca Lora 4bit
Made some adjust for the code in peft and gptq for llama, and make it possible for lora finetuning with a 4 bits base model. The same adjustment can be made for 2, 3 and 8 bits.
<br>
~Still numerically unstable.~ Resolved.
<br>
Reconstruct fp16 matrix from 4bit data and call torch.matmul largely increased the inference speed.
<br>
Added install script for windows and linux.
<br>

# Requirements
gptq-for-llama: https://github.com/qwopqwop200/GPTQ-for-LLaMa<br>
peft: https://github.com/huggingface/peft.git<br>
<br>

# Install
~copy files from GPTQ-for-LLaMa into GPTQ-for-LLaMa path and re-compile cuda extension~<br>
~copy files from peft/tuners/lora.py to peft path, replace it~<br>

Linux:

```
./install.sh
```

Windows:

```
./install.bat
```

# Finetune
~The same finetune script from https://github.com/tloen/alpaca-lora can be used.~<br>

After installation, this script can be used:

```
python finetune.py
```

# Inference

After installation, this script can be used:

```
python inference.py
```