|
--- |
|
language: |
|
- en |
|
--- |
|
## Information |
|
|
|
This is a Exl2 quantized version of [MN-12B-Starsong-v1](https://huggingface.co/aetherwiing/MN-12B-Starsong-v1) |
|
|
|
Please refer to the original creator for more information. |
|
|
|
Calibration dataset: Exl2 default |
|
|
|
## Branches: |
|
|
|
- main: Measurement files |
|
- 4bpw: 4 bits per weight |
|
- 5bpw: 5 bits per weight |
|
- 6bpw: 6 bits per weight |
|
|
|
## Notes |
|
|
|
- 6bpw is recommended for the best quality to vram usage ratio (assuming you have enough vram). |
|
- Quants greater than 6bpw will not be created because there is no improvement in using them. If you really want them, ask someone else or make them yourself. |
|
|
|
|
|
## Download |
|
|
|
|
|
With [async-hf-downloader](https://github.com/theroyallab/async-hf-downloader): A lightweight and asynchronous huggingface downloader created by me |
|
|
|
|
|
```shell |
|
./async-hf-downloader royallab/MN-12B-Starsong-v1-exl2 -r 6bpw -p MN-12B-Starsong-v1-exl2-6bpw |
|
``` |
|
|
|
|
|
With HuggingFace hub (`pip install huggingface_hub`) |
|
|
|
```shell |
|
huggingface-cli download royallab/MN-12B-Starsong-v1-exl2 --revision 6bpw --local-dir MN-12B-Starsong-v1-exl2-6bpw |
|
``` |
|
|
|
## Run in TabbyAPI |
|
|
|
TabbyAPI is a pure exllamav2 FastAPI server developed by us. You can find TabbyAPI's source code here: [https://github.com/theroyallab/TabbyAPI](https://github.com/theroyallab/TabbyAPI) |
|
|
|
1. Inside TabbyAPI's config.yml, set `model_name` to `MN-12B-Starsong-v1-exl2-6bpw` |
|
|
|
1. You can also use an argument `--model_name MN-12B-Starsong-v1-exl2-6bpw` on startup or you can use the `/v1/model/load` endpoint |
|
|
|
2. Launch TabbyAPI inside your python env by running `./start.bat` or `./start.sh` |
|
|
|
|
|
## Donate? |
|
|
|
All my infrastructure and cloud expenses are paid out of pocket. If you'd like to donate, you can do so here: https://ko-fi.com/kingbri |
|
|
|
You should not feel obligated to donate, but if you do, I'd appreciate it. |
|
--- |
|
|
