README.md
5.5 KB · 97 lines · markdown Raw
1 ---
2 base_model: pixas/Miner-8B
3 datasets:
4 - agentica-org/DeepScaleR-Preview-Dataset
5 language:
6 - en
7 library_name: transformers
8 license: apache-2.0
9 mradermacher:
10 readme_rev: 1
11 quantized_by: mradermacher
12 tags:
13 - transformers
14 - reasoning
15 - reinforcement-learning
16 - rlvr
17 - math
18 - miner
19 - qwen3
20 - causal-lm
21 ---
22 ## About
23
24 <!-- ### quantize_version: 2 -->
25 <!-- ### output_tensor_quantised: 1 -->
26 <!-- ### convert_type: hf -->
27 <!-- ### vocab_type: -->
28 <!-- ### tags: nicoboss -->
29 <!-- ### quants: Q2_K IQ3_M Q4_K_S IQ3_XXS Q3_K_M small-IQ4_NL Q4_K_M IQ2_M Q6_K IQ4_XS Q2_K_S IQ1_M Q3_K_S IQ2_XXS Q3_K_L IQ2_XS Q5_K_S IQ2_S IQ1_S Q5_K_M Q4_0 IQ3_XS Q4_1 IQ3_S -->
30 <!-- ### quants_skip: -->
31 <!-- ### skip_mmproj: -->
32 weighted/imatrix quants of https://huggingface.co/pixas/Miner-8B
33
34 <!-- provided-files -->
35
36 ***For a convenient overview and download list, visit our [model page for this model](https://hf.tst.eu/model#Miner-8B-i1-GGUF).***
37
38 static quants are available at https://huggingface.co/mradermacher/Miner-8B-GGUF
39 ## Usage
40
41 If you are unsure how to use GGUF files, refer to one of [TheBloke's
42 READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
43 more details, including on how to concatenate multi-part files.
44
45 ## Provided Quants
46
47 (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
48
49 | Link | Type | Size/GB | Notes |
50 |:-----|:-----|--------:|:------|
51 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.imatrix.gguf) | imatrix | 0.1 | imatrix file (for creating your own quants) |
52 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ1_S.gguf) | i1-IQ1_S | 2.2 | for the desperate |
53 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ1_M.gguf) | i1-IQ1_M | 2.4 | mostly desperate |
54 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 2.6 | |
55 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ2_XS.gguf) | i1-IQ2_XS | 2.8 | |
56 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ2_S.gguf) | i1-IQ2_S | 3.0 | |
57 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ2_M.gguf) | i1-IQ2_M | 3.2 | |
58 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q2_K_S.gguf) | i1-Q2_K_S | 3.2 | very low quality |
59 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q2_K.gguf) | i1-Q2_K | 3.4 | IQ3_XXS probably better |
60 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 3.5 | lower quality |
61 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ3_XS.gguf) | i1-IQ3_XS | 3.7 | |
62 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q3_K_S.gguf) | i1-Q3_K_S | 3.9 | IQ3_XS probably better |
63 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ3_S.gguf) | i1-IQ3_S | 3.9 | beats Q3_K* |
64 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ3_M.gguf) | i1-IQ3_M | 4.0 | |
65 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q3_K_M.gguf) | i1-Q3_K_M | 4.2 | IQ3_S probably better |
66 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q3_K_L.gguf) | i1-Q3_K_L | 4.5 | IQ3_M probably better |
67 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ4_XS.gguf) | i1-IQ4_XS | 4.7 | |
68 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q4_0.gguf) | i1-Q4_0 | 4.9 | fast, low quality |
69 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-IQ4_NL.gguf) | i1-IQ4_NL | 4.9 | prefer IQ4_XS |
70 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q4_K_S.gguf) | i1-Q4_K_S | 4.9 | optimal size/speed/quality |
71 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q4_K_M.gguf) | i1-Q4_K_M | 5.1 | fast, recommended |
72 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q4_1.gguf) | i1-Q4_1 | 5.3 | |
73 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q5_K_S.gguf) | i1-Q5_K_S | 5.8 | |
74 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q5_K_M.gguf) | i1-Q5_K_M | 6.0 | |
75 | [GGUF](https://huggingface.co/mradermacher/Miner-8B-i1-GGUF/resolve/main/Miner-8B.i1-Q6_K.gguf) | i1-Q6_K | 6.8 | practically like static Q6_K |
76
77 Here is a handy graph by ikawrakow comparing some lower-quality quant
78 types (lower is better):
79
80 ![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)
81
82 And here are Artefact2's thoughts on the matter:
83 https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
84
85 ## FAQ / Model Request
86
87 See https://huggingface.co/mradermacher/model_requests for some answers to
88 questions you might have and/or if you want some other model quantized.
89
90 ## Thanks
91
92 I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
93 me use its servers and providing upgrades to my workstation to enable
94 this work in my free time. Additional thanks to [@nicoboss](https://huggingface.co/nicoboss) for giving me access to his private supercomputer, enabling me to provide many more imatrix quants, at much higher quality, than I would otherwise be able to.
95
96 <!-- end -->
97