File size: 3,668 Bytes
12b2fde f47aca5 12b2fde |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 |
---
datasets:
- ewof/koishi-instruct-metharme
exported_from: ewof/koishi-8x7b-qlora
language:
- en
library_name: transformers
quantized_by: mradermacher
---
## About
<!-- ### quantize_version: 1 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: -->
<!-- ### vocab_type: -->
weighted/imatrix quants of https://huggingface.co/ewof/koishi-8x7b-qlora
<!-- provided-files -->
static quants are available at https://huggingface.co/mradermacher/koishi-8x7b-qlora-GGUF
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-IQ2_M.gguf) | i1-IQ2_M | 15.6 | |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-Q2_K.gguf) | i1-Q2_K | 17.4 | IQ3_XXS probably better |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 18.3 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-IQ3_XS.gguf) | i1-IQ3_XS | 19.5 | |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-IQ3_S.gguf) | i1-IQ3_S | 20.5 | beats Q3_K* |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-Q3_K_S.gguf) | i1-Q3_K_S | 20.5 | IQ3_XS probably better |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-IQ3_M.gguf) | i1-IQ3_M | 21.5 | |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-Q3_K_M.gguf) | i1-Q3_K_M | 22.6 | IQ3_S probably better |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-Q3_K_L.gguf) | i1-Q3_K_L | 24.3 | IQ3_M probably better |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-IQ4_XS.gguf) | i1-IQ4_XS | 25.2 | |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-Q4_0.gguf) | i1-Q4_0 | 26.7 | fast, low quality |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-Q4_K_M.gguf) | i1-Q4_K_M | 28.5 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-Q5_K_S.gguf) | i1-Q5_K_S | 32.3 | |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-Q5_K_M.gguf) | i1-Q5_K_M | 33.3 | |
| [GGUF](https://huggingface.co/mradermacher/koishi-8x7b-qlora-i1-GGUF/resolve/main/koishi-8x7b-qlora.i1-Q6_K.gguf) | i1-Q6_K | 38.5 | practically like static Q6_K |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time.
<!-- end -->
|