mradermacher commited on
Commit
271948a
1 Parent(s): b04e0e7

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +67 -0
README.md CHANGED
@@ -1,5 +1,72 @@
 
 
 
 
 
 
 
 
 
 
1
  <!-- ### quantize_version: 2 -->
2
  <!-- ### output_tensor_quantised: 1 -->
3
  <!-- ### convert_type: hf -->
4
  <!-- ### vocab_type: -->
5
  weighted/imatrix quants of https://huggingface.co/cloudyu/Llama-3-70Bx2-MOE
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: cloudyu/Llama-3-70Bx2-MOE
3
+ language:
4
+ - en
5
+ library_name: transformers
6
+ license: llama3
7
+ quantized_by: mradermacher
8
+ ---
9
+ ## About
10
+
11
  <!-- ### quantize_version: 2 -->
12
  <!-- ### output_tensor_quantised: 1 -->
13
  <!-- ### convert_type: hf -->
14
  <!-- ### vocab_type: -->
15
  weighted/imatrix quants of https://huggingface.co/cloudyu/Llama-3-70Bx2-MOE
16
+
17
+ <!-- provided-files -->
18
+ static quants are available at https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-GGUF
19
+ ## Usage
20
+
21
+ If you are unsure how to use GGUF files, refer to one of [TheBloke's
22
+ READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
23
+ more details, including on how to concatenate multi-part files.
24
+
25
+ ## Provided Quants
26
+
27
+ (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
28
+
29
+ | Link | Type | Size/GB | Notes |
30
+ |:-----|:-----|--------:|:------|
31
+ | [GGUF](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ1_S.gguf) | i1-IQ1_S | 26.8 | for the desperate |
32
+ | [GGUF](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ1_M.gguf) | i1-IQ1_M | 29.4 | mostly desperate |
33
+ | [GGUF](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 33.9 | |
34
+ | [GGUF](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ2_XS.gguf) | i1-IQ2_XS | 37.6 | |
35
+ | [GGUF](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ2_S.gguf) | i1-IQ2_S | 39.0 | |
36
+ | [GGUF](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ2_M.gguf) | i1-IQ2_M | 42.5 | |
37
+ | [GGUF](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q2_K.gguf) | i1-Q2_K | 46.9 | IQ3_XXS probably better |
38
+ | [GGUF](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 49.2 | lower quality |
39
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ3_XS.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ3_XS.gguf.part2of2) | i1-IQ3_XS | 52.3 | |
40
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ3_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ3_S.gguf.part2of2) | i1-IQ3_S | 55.2 | beats Q3_K* |
41
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q3_K_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q3_K_S.gguf.part2of2) | i1-Q3_K_S | 55.2 | IQ3_XS probably better |
42
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ3_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ3_M.gguf.part2of2) | i1-IQ3_M | 56.6 | |
43
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q3_K_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q3_K_M.gguf.part2of2) | i1-Q3_K_M | 61.2 | IQ3_S probably better |
44
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q3_K_L.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q3_K_L.gguf.part2of2) | i1-Q3_K_L | 66.3 | IQ3_M probably better |
45
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ4_XS.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-IQ4_XS.gguf.part2of2) | i1-IQ4_XS | 68.0 | |
46
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q4_0.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q4_0.gguf.part2of2) | i1-Q4_0 | 72.1 | fast, low quality |
47
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q4_K_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q4_K_S.gguf.part2of2) | i1-Q4_K_S | 72.5 | optimal size/speed/quality |
48
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q4_K_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q4_K_M.gguf.part2of2) | i1-Q4_K_M | 76.8 | fast, recommended |
49
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q5_K_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q5_K_S.gguf.part2of2) | i1-Q5_K_S | 87.5 | |
50
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q5_K_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q5_K_M.gguf.part2of2) | i1-Q5_K_M | 90.1 | |
51
+ | [PART 1](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q6_K.gguf.part1of3) [PART 2](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q6_K.gguf.part2of3) [PART 3](https://huggingface.co/mradermacher/Llama-3-70Bx2-MOE-i1-GGUF/resolve/main/Llama-3-70Bx2-MOE.i1-Q6_K.gguf.part3of3) | i1-Q6_K | 104.2 | practically like static Q6_K |
52
+
53
+ Here is a handy graph by ikawrakow comparing some lower-quality quant
54
+ types (lower is better):
55
+
56
+ ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
57
+
58
+ And here are Artefact2's thoughts on the matter:
59
+ https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
60
+
61
+ ## FAQ / Model Request
62
+
63
+ See https://huggingface.co/mradermacher/model_requests for some answers to
64
+ questions you might have and/or if you want some other model quantized.
65
+
66
+ ## Thanks
67
+
68
+ I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
69
+ me use its servers and providing upgrades to my workstation to enable
70
+ this work in my free time.
71
+
72
+ <!-- end -->