mradermacher commited on
Commit
9e71bf5
1 Parent(s): cc87f52

auto-patch README.md

Browse files
Files changed (1) hide show
  1. README.md +70 -0
README.md CHANGED
@@ -1,5 +1,75 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  <!-- ### quantize_version: 1 -->
2
  <!-- ### output_tensor_quantised: 1 -->
3
  <!-- ### convert_type: -->
4
  <!-- ### vocab_type: -->
5
  weighted/imatrix quants of https://huggingface.co/Dogge/Tia-70B-RP
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: unsloth/llama-3-70b-Instruct-bnb-4bit
3
+ exported_from: Dogge/Tia-70B-RP
4
+ language:
5
+ - en
6
+ library_name: transformers
7
+ license: apache-2.0
8
+ quantized_by: mradermacher
9
+ tags:
10
+ - text-generation-inference
11
+ - transformers
12
+ - unsloth
13
+ - llama
14
+ - trl
15
+ - sft
16
+ ---
17
+ ## About
18
+
19
  <!-- ### quantize_version: 1 -->
20
  <!-- ### output_tensor_quantised: 1 -->
21
  <!-- ### convert_type: -->
22
  <!-- ### vocab_type: -->
23
  weighted/imatrix quants of https://huggingface.co/Dogge/Tia-70B-RP
24
+
25
+
26
+ <!-- provided-files -->
27
+ static quants are available at https://huggingface.co/mradermacher/Tia-70B-RP-GGUF
28
+ ## Usage
29
+
30
+ If you are unsure how to use GGUF files, refer to one of [TheBloke's
31
+ READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
32
+ more details, including on how to concatenate multi-part files.
33
+
34
+ ## Provided Quants
35
+
36
+ (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
37
+
38
+ | Link | Type | Size/GB | Notes |
39
+ |:-----|:-----|--------:|:------|
40
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 19.2 | |
41
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-IQ2_XS.gguf) | i1-IQ2_XS | 21.2 | |
42
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-IQ2_S.gguf) | i1-IQ2_S | 22.3 | |
43
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-IQ2_M.gguf) | i1-IQ2_M | 24.2 | |
44
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q2_K.gguf) | i1-Q2_K | 26.5 | IQ3_XXS probably better |
45
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 27.6 | lower quality |
46
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-IQ3_XS.gguf) | i1-IQ3_XS | 29.4 | |
47
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-IQ3_S.gguf) | i1-IQ3_S | 31.0 | beats Q3_K* |
48
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q3_K_S.gguf) | i1-Q3_K_S | 31.0 | IQ3_XS probably better |
49
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-IQ3_M.gguf) | i1-IQ3_M | 32.0 | |
50
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q3_K_M.gguf) | i1-Q3_K_M | 34.4 | IQ3_S probably better |
51
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q3_K_L.gguf) | i1-Q3_K_L | 37.2 | IQ3_M probably better |
52
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-IQ4_XS.gguf) | i1-IQ4_XS | 38.0 | |
53
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q4_0.gguf) | i1-Q4_0 | 40.2 | fast, low quality |
54
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q4_K_S.gguf) | i1-Q4_K_S | 40.4 | optimal size/speed/quality |
55
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q4_K_M.gguf) | i1-Q4_K_M | 42.6 | fast, recommended |
56
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q5_K_S.gguf) | i1-Q5_K_S | 48.8 | |
57
+ | [GGUF](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q5_K_M.gguf) | i1-Q5_K_M | 50.1 | |
58
+ | [PART 1](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q6_K.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Tia-70B-RP-i1-GGUF/resolve/main/Tia-70B-RP.i1-Q6_K.gguf.part2of2) | i1-Q6_K | 58.0 | practically like static Q6_K |
59
+
60
+
61
+ Here is a handy graph by ikawrakow comparing some lower-quality quant
62
+ types (lower is better):
63
+
64
+ ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
65
+
66
+ And here are Artefact2's thoughts on the matter:
67
+ https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
68
+
69
+ ## Thanks
70
+
71
+ I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
72
+ me use its servers and providing upgrades to my workstation to enable
73
+ this work in my free time.
74
+
75
+ <!-- end -->