File size: 2,846 Bytes
c439873 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 |
---
quantized_by: bartowski
pipeline_tag: text-generation
---
## Exllama v2 Quantizations of Llama-3-Instruct-8B-SimPO
Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.1.0">turboderp's ExLlamaV2 v0.1.0</a> for quantization.
<b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
Original model: https://huggingface.co/princeton-nlp/Llama-3-Instruct-8B-SimPO
## Prompt format
No chat template specified so default is used. This may be incorrect, check original model card for details.
```
<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
```
## Available sizes
| Branch | Bits | lm_head bits | VRAM (4k) | VRAM (8K) | VRAM (16k) | VRAM (32k) | Description |
| ----- | ---- | ------- | ------ | ------ | ------ | ------ | ------------ |
| [8_0](https://huggingface.co/bartowski/Llama-3-Instruct-8B-SimPO-exl2/tree/8_0) | 8.0 | 8.0 | 10.1 GB | 10.5 GB | 11.5 GB | 13.6 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
| [6_5](https://huggingface.co/bartowski/Llama-3-Instruct-8B-SimPO-exl2/tree/6_5) | 6.5 | 8.0 | 8.9 GB | 9.3 GB | 10.3 GB | 12.4 GB | Very similar to 8.0, good tradeoff of size vs performance, **recommended**. |
| [5_0](https://huggingface.co/bartowski/Llama-3-Instruct-8B-SimPO-exl2/tree/5_0) | 5.0 | 6.0 | 7.7 GB | 8.1 GB | 9.1 GB | 11.2 GB | Slightly lower quality vs 6.5, but usable on 8GB cards. |
| [4_25](https://huggingface.co/bartowski/Llama-3-Instruct-8B-SimPO-exl2/tree/4_25) | 4.25 | 6.0 | 7.0 GB | 7.4 GB | 8.4 GB | 10.5 GB | GPTQ equivalent bits per weight, slightly higher quality. |
| [3_5](https://huggingface.co/bartowski/Llama-3-Instruct-8B-SimPO-exl2/tree/3_5) | 3.5 | 6.0 | 6.4 GB | 6.8 GB | 7.8 GB | 9.9 GB | Lower quality, only use if you have to. |
## Download instructions
With git:
```shell
git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/Llama-3-Instruct-8B-SimPO-exl2 Llama-3-Instruct-8B-SimPO-exl2-6_5
```
With huggingface hub (credit to TheBloke for instructions):
```shell
pip3 install huggingface-hub
```
To download a specific branch, use the `--revision` parameter. For example, to download the 6.5 bpw branch:
Linux:
```shell
huggingface-cli download bartowski/Llama-3-Instruct-8B-SimPO-exl2 --revision 6_5 --local-dir Llama-3-Instruct-8B-SimPO-exl2-6_5
```
Windows (which apparently doesn't like _ in folders sometimes?):
```shell
huggingface-cli download bartowski/Llama-3-Instruct-8B-SimPO-exl2 --revision 6_5 --local-dir Llama-3-Instruct-8B-SimPO-exl2-6.5
```
Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski
|