Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,35 @@
|
|
1 |
---
|
2 |
license: cc-by-sa-4.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: cc-by-sa-4.0
|
3 |
+
datasets:
|
4 |
+
- mc4
|
5 |
+
- cc100
|
6 |
+
- oscar
|
7 |
+
- togethercomputer/RedPajama-Data-1T
|
8 |
+
language:
|
9 |
+
- ja
|
10 |
+
- en
|
11 |
+
base_model: meta-llama/Llama-2-70b-hf
|
12 |
+
pipeline_tag: text-generation
|
13 |
+
tags:
|
14 |
+
- llama
|
15 |
+
- llama-2
|
16 |
---
|
17 |
+
|
18 |
+
This repository contains the exl2 quants of [karakuri-ai/karakuri-lm-70b-chat-v0.1](https://huggingface.co/karakuri-ai/karakuri-lm-70b-chat-v0.1), calibrated using [the default dataset built by exllamav2](https://github.com/turboderp/exllamav2/blob/3b0f5230e9619fc0ddf1f302785049d10a65fe26/conversion/tokenize.py#L55).
|
19 |
+
|
20 |
+
Compatible with [exllamav2](https://github.com/turboderp/exllamav2) version 0.0.11 and later. For optimal model loading, it is recommended to use [tabbyAPI](https://github.com/theroyallab/tabbyAPI).
|
21 |
+
|
22 |
+
The measurement file is attached in the branch `main` and all quants are stored in their respective branches.
|
23 |
+
|
24 |
+
The chart below presents the calibration perplexity and [wikitext-103-v1](https://huggingface.co/datasets/wikitext) test perplexity for all provided quants.
|
25 |
+
|
26 |
+
| Quant | Calibration Perplexity | wikitext-103-v1 Test Perplexity |
|
27 |
+
|-----------------------------------------------------------------------------------------|------------------------|---------------------------------|
|
28 |
+
| [2.4bpw-h6](https://huggingface.co/MatrixC7/karakuri-lm-70b-chat-v0.1-exl2/tree/2.4bpw-h6) | 8.3509 | 7.1054 |
|
29 |
+
| [2.65bpw-h6](https://huggingface.co/MatrixC7/karakuri-lm-70b-chat-v0.1-exl2/tree/2.65bpw-h6) | 8.0491 | 6.4045 |
|
30 |
+
| [3.0bpw-h6](https://huggingface.co/MatrixC7/karakuri-lm-70b-chat-v0.1-exl2/tree/3.0bpw-h6) | 7.7330 | 6.0661 |
|
31 |
+
| [4.0bpw-h6](https://huggingface.co/MatrixC7/karakuri-lm-70b-chat-v0.1-exl2/tree/4.0bpw-h6) | 7.4540 | 5.6886 |
|
32 |
+
| [4.65bpw-h6](https://huggingface.co/MatrixC7/karakuri-lm-70b-chat-v0.1-exl2/tree/4.65bpw-h6) | 7.4089 | 5.5980 |
|
33 |
+
| [5.0bpw-h6](https://huggingface.co/MatrixC7/karakuri-lm-70b-chat-v0.1-exl2/tree/5.0bpw-h6) | 7.3924 | |
|
34 |
+
| [6.0bpw-h6](https://huggingface.co/MatrixC7/karakuri-lm-70b-chat-v0.1-exl2/tree/6.0bpw-h6) | 7.3674 | |
|
35 |
+
| [8.0bpw-h8](https://huggingface.co/MatrixC7/karakuri-lm-70b-chat-v0.1-exl2/tree/8.0bpw-h8) | 7.3603 | |
|