---
license: other
inference: false
---

# Quantised GGMLs of alpaca-lora-65B

Quantised 2bit, 4bit and 5bit GGMLs of [changsung's alpaca-lora-65B](https://huggingface.co/chansung/alpaca-lora-65b) for CPU inference with [llama.cpp](https://github.com/ggerganov/llama.cpp).

I also have 4bit GPTQ files for GPU inference available here: [TheBloke/alpaca-lora-65B-GPTQ-4bit](https://huggingface.co/TheBloke/alpaca-lora-65B-GPTQ-4bit).

## Provided files
| Name | Quant method | Bits | Size | RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
`alpaca-lora-65B.ggml.q2_0.bin` | q2_0 | 2bit | 24.5GB | 27GB | Lowest RAM requirements, minimum quality | 
`alpaca-lora-65B.ggml.q4_0.bin` | q4_0 | 4bit | 40.8GB | 43GB | Maximum compatibility |
`alpaca-lora-65B.ggml.q4_2.bin` | q4_2 | 4bit | 40.8GB | 43GB | Best compromise between resources, speed and quality |
`alpaca-lora-65B.ggml.q5_0.bin` | q5_0 | 5bit | 44.9GB | 47GB | Brand new 5bit method. Potentially higher quality than 4bit, at cost of slightly higher resources. |
`alpaca-lora-65B.ggml.q5_1.bin` | q5_1 | 5bit | 49GB | 51GB | Brand new 5bit method. Slightly higher resource usage than q5_0. |

* The q2_0 file requires the least resources, but does not have great quality compared to the others.
  * It's likely to be better to use a 30B model at 4bit vs a 65B model at 2bit.
* The q4_0 file provides lower quality, but maximal compatibility. It will work with past and future versions of llama.cpp
* The q4_2 file offers the best combination of performance and quality. This format is still subject to change and there may be compatibility issues, see below.
* The q5_0 file is using brand new 5bit method released 26th April. This is the 5bit equivalent of q4_0.
* The q5_1 file is using brand new 5bit method released 26th April. This is the 5bit equivalent of q4_1.

## q4_2 compatibility

q4_2 is a relatively new 4bit quantisation method offering improved quality. However they are still under development and their formats are subject to change.

In order to use these files you will need to use recent llama.cpp code. And it's possible that future updates to llama.cpp could require that these files are re-generated.

If and when the q4_2 file no longer works with recent versions of llama.cpp I will endeavour to update it.

If you want to ensure guaranteed compatibility with a wide range of llama.cpp versions, use the q4_0 file.

## q5_0 and q5_1 compatibility

These new methods were released to llama.cpp on 26th April. You will need to pull the latest llama.cpp code and rebuild to be able to use them.

Don't expect any third-party UIs/tools to support them yet.

### 2bit q2_0 compatibility

This file was created using an experimental 2bit method being trialled in [llama.cpp PR 1004](https://github.com/ggerganov/llama.cpp/pull/1004).

This code is not yet merged into the main `llama.cpp` repo and it is not clear if it ever will be.

To run this file you need to compile and run the same `llama.cpp` code that was used to create it.

To checkout this code and compile this version, do the following:
```
git clone https://github.com/sw/llama.cpp llama-q2q3
cd llama-q2q3
git checkout q2q3
make
```

## How to run in `llama.cpp`

I use the following command line; adjust for your tastes and needs:

```
./main -t 18 -m alpaca-lora-65B.ggml.q4_2.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
Write a story about llamas
### Response:"
```
Change `-t 18` to the number of physical CPU cores you have. For example if your system has 8 cores/16 threads, use `-t 8`.

If you want to have a chat-style conversation, replace the `-p <PROMPT>` argument with `-i -ins`

## How to run in `text-generation-webui`

Further instructions here: [text-generation-webui/docs/llama.cpp-models.md](https://github.com/oobabooga/text-generation-webui/blob/main/docs/llama.cpp-models.md).

Note: at this time text-generation-webui will not support the new q5 quantisation methods.

**Thireus** has written a [great guide on how to update it to the latest llama.cpp code](https://huggingface.co/TheBloke/wizardLM-7B-GGML/discussions/5) so that these files can be used in the UI.

# Original model card not provided

No model card was provided in [changsung's original repository](https://huggingface.co/chansung/alpaca-lora-65b).

Based on the name, I assume this is the result of fine tuning using the original GPT 3.5 Alpaca dataset. It is unknown as to whether the original Stanford data was used, or the [cleaned tloen/alpaca-lora variant](https://github.com/tloen/alpaca-lora).