mradermacher
commited on
Commit
•
1fc25ab
1
Parent(s):
d422d67
auto-patch README.md
Browse files
README.md
CHANGED
@@ -1,5 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
<!-- ### quantize_version: 1 -->
|
2 |
<!-- ### output_tensor_quantised: 1 -->
|
3 |
<!-- ### convert_type: -->
|
4 |
<!-- ### vocab_type: -->
|
5 |
weighted/imatrix quants of https://huggingface.co/ibivibiv/strix-rufipes-70b
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
exported_from: ibivibiv/strix-rufipes-70b
|
3 |
+
language:
|
4 |
+
- en
|
5 |
+
library_name: transformers
|
6 |
+
license: llama2
|
7 |
+
quantized_by: mradermacher
|
8 |
+
tags:
|
9 |
+
- logic
|
10 |
+
- planning
|
11 |
+
---
|
12 |
+
## About
|
13 |
+
|
14 |
<!-- ### quantize_version: 1 -->
|
15 |
<!-- ### output_tensor_quantised: 1 -->
|
16 |
<!-- ### convert_type: -->
|
17 |
<!-- ### vocab_type: -->
|
18 |
weighted/imatrix quants of https://huggingface.co/ibivibiv/strix-rufipes-70b
|
19 |
+
|
20 |
+
|
21 |
+
<!-- provided-files -->
|
22 |
+
static quants are available at https://huggingface.co/mradermacher/strix-rufipes-70b-GGUF
|
23 |
+
## Usage
|
24 |
+
|
25 |
+
If you are unsure how to use GGUF files, refer to one of [TheBloke's
|
26 |
+
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
|
27 |
+
more details, including on how to concatenate multi-part files.
|
28 |
+
|
29 |
+
## Provided Quants
|
30 |
+
|
31 |
+
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
|
32 |
+
|
33 |
+
| Link | Type | Size/GB | Notes |
|
34 |
+
|:-----|:-----|--------:|:------|
|
35 |
+
| [GGUF](https://huggingface.co/mradermacher/strix-rufipes-70b-i1-GGUF/resolve/main/strix-rufipes-70b.i1-Q2_K.gguf) | i1-Q2_K | 25.6 | IQ3_XXS probably better |
|
36 |
+
| [GGUF](https://huggingface.co/mradermacher/strix-rufipes-70b-i1-GGUF/resolve/main/strix-rufipes-70b.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 26.7 | lower quality |
|
37 |
+
| [GGUF](https://huggingface.co/mradermacher/strix-rufipes-70b-i1-GGUF/resolve/main/strix-rufipes-70b.i1-Q3_K_M.gguf) | i1-Q3_K_M | 33.4 | IQ3_S probably better |
|
38 |
+
| [GGUF](https://huggingface.co/mradermacher/strix-rufipes-70b-i1-GGUF/resolve/main/strix-rufipes-70b.i1-Q4_K_S.gguf) | i1-Q4_K_S | 39.3 | optimal size/speed/quality |
|
39 |
+
| [GGUF](https://huggingface.co/mradermacher/strix-rufipes-70b-i1-GGUF/resolve/main/strix-rufipes-70b.i1-Q4_K_M.gguf) | i1-Q4_K_M | 41.5 | fast, recommended |
|
40 |
+
|
41 |
+
|
42 |
+
Here is a handy graph by ikawrakow comparing some lower-quality quant
|
43 |
+
types (lower is better):
|
44 |
+
|
45 |
+
![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
|
46 |
+
|
47 |
+
And here are Artefact2's thoughts on the matter:
|
48 |
+
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
|
49 |
+
|
50 |
+
## Thanks
|
51 |
+
|
52 |
+
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
|
53 |
+
me use its servers and providing upgrades to my workstation to enable
|
54 |
+
this work in my free time.
|
55 |
+
|
56 |
+
<!-- end -->
|