measurement.json
Browse files- README.md +96 -0
- measurement.json +0 -0
README.md
ADDED
@@ -0,0 +1,96 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: other
|
3 |
+
license_name: llama3
|
4 |
+
license_link: https://huggingface.co/meta-llama/Meta-Llama-3-8B/blob/main/LICENSE
|
5 |
+
base_model: meta-llama/Meta-Llama-3-8B
|
6 |
+
tags:
|
7 |
+
- llama-3
|
8 |
+
datasets:
|
9 |
+
- jondurbin/airoboros-3.2
|
10 |
+
- bluemoon-fandom-1-1-rp-cleaned
|
11 |
+
- boolq
|
12 |
+
- LDJnr/Capybara
|
13 |
+
- jondurbin/cinematika-v0.1
|
14 |
+
- glaiveai/glaive-function-calling-v2
|
15 |
+
- grimulkan/LimaRP-augmented
|
16 |
+
- piqa
|
17 |
+
- Vezora/Tested-22k-Python-Alpaca
|
18 |
+
- mattpscott/airoboros-summarization
|
19 |
+
- unalignment/toxic-dpo-v0.2
|
20 |
+
- allenai/ultrafeedback_binarized_cleaned
|
21 |
+
- argilla/distilabel-intel-orca-dpo-pairs
|
22 |
+
- jondurbin/airoboros-3.2
|
23 |
+
- jondurbin/contextual-dpo-v0.1
|
24 |
+
- jondurbin/gutenberg-dpo-v0.1
|
25 |
+
- jondurbin/py-dpo-v0.1
|
26 |
+
- jondurbin/truthy-dpo-v0.1
|
27 |
+
- jondurbin/gutenberg-dpo-v0.1
|
28 |
+
- lmsys/lmsys-chat-1m
|
29 |
+
quantized_by: bartowski
|
30 |
+
pipeline_tag: text-generation
|
31 |
+
---
|
32 |
+
|
33 |
+
## Exllama v2 Quantizations of airoboros-dpo-70b-3.3
|
34 |
+
|
35 |
+
Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.21">turboderp's ExLlamaV2 v0.0.21</a> for quantization.
|
36 |
+
|
37 |
+
<b>The "main" branch only contains the measurement.json, download one of the other branches for the model (see below)</b>
|
38 |
+
|
39 |
+
Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
|
40 |
+
|
41 |
+
Conversion was done using the default calibration dataset.
|
42 |
+
|
43 |
+
Default arguments used except when the bits per weight is above 6.0, at that point the lm_head layer is quantized at 8 bits per weight instead of the default 6.
|
44 |
+
|
45 |
+
Original model: https://huggingface.co/jondurbin/airoboros-dpo-70b-3.3
|
46 |
+
|
47 |
+
|
48 |
+
<a href="https://huggingface.co/bartowski/airoboros-dpo-70b-3.3-exl2/tree/6_5">6.5 bits per weight</a>
|
49 |
+
|
50 |
+
<a href="https://huggingface.co/bartowski/airoboros-dpo-70b-3.3-exl2/tree/4_25">4.25 bits per weight</a>
|
51 |
+
|
52 |
+
<a href="https://huggingface.co/bartowski/airoboros-dpo-70b-3.3-exl2/tree/3_5">3.5 bits per weight</a>
|
53 |
+
|
54 |
+
<a href="https://huggingface.co/bartowski/airoboros-dpo-70b-3.3-exl2/tree/3_0">3.0 bits per weight</a>
|
55 |
+
|
56 |
+
<a href="https://huggingface.co/bartowski/airoboros-dpo-70b-3.3-exl2/tree/2_5">2.5 bits per weight</a>
|
57 |
+
|
58 |
+
<a href="https://huggingface.co/bartowski/airoboros-dpo-70b-3.3-exl2/tree/2_2">2.2 bits per weight</a>
|
59 |
+
|
60 |
+
|
61 |
+
## Download instructions
|
62 |
+
|
63 |
+
With git:
|
64 |
+
|
65 |
+
```shell
|
66 |
+
git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/airoboros-dpo-70b-3.3-exl2
|
67 |
+
```
|
68 |
+
|
69 |
+
With huggingface hub (credit to TheBloke for instructions):
|
70 |
+
|
71 |
+
```shell
|
72 |
+
pip3 install huggingface-hub
|
73 |
+
```
|
74 |
+
|
75 |
+
To download the `main` (only useful if you only care about measurement.json) branch to a folder called `airoboros-dpo-70b-3.3-exl2`:
|
76 |
+
|
77 |
+
```shell
|
78 |
+
mkdir airoboros-dpo-70b-3.3-exl2
|
79 |
+
huggingface-cli download bartowski/airoboros-dpo-70b-3.3-exl2 --local-dir airoboros-dpo-70b-3.3-exl2 --local-dir-use-symlinks False
|
80 |
+
```
|
81 |
+
|
82 |
+
To download from a different branch, add the `--revision` parameter:
|
83 |
+
|
84 |
+
Linux:
|
85 |
+
|
86 |
+
```shell
|
87 |
+
mkdir airoboros-dpo-70b-3.3-exl2-6_5
|
88 |
+
huggingface-cli download bartowski/airoboros-dpo-70b-3.3-exl2 --revision 6_5 --local-dir airoboros-dpo-70b-3.3-exl2-6_5 --local-dir-use-symlinks False
|
89 |
+
```
|
90 |
+
|
91 |
+
Windows (which apparently doesn't like _ in folders sometimes?):
|
92 |
+
|
93 |
+
```shell
|
94 |
+
mkdir airoboros-dpo-70b-3.3-exl2-6.5
|
95 |
+
huggingface-cli download bartowski/airoboros-dpo-70b-3.3-exl2 --revision 6_5 --local-dir airoboros-dpo-70b-3.3-exl2-6.5 --local-dir-use-symlinks False
|
96 |
+
```
|
measurement.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|