README.md

---
exported_from: davidkim205/Rhea-72b-v0.5
language:
- en
library_name: transformers
license: apache-2.0
quantized_by: mradermacher
---
## About

weighted/imatrix quants of https://huggingface.co/davidkim205/Rhea-72b-v0.5

**the imatrix was calculated on a reduced 40k token set (the "quarter" set) as the full token set caused overflows in the model (likely a model bug)**

<!-- provided-files -->
static quants are available at https://huggingface.co/mradermacher/Rhea-72b-v0.5-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/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-IQ2_M.gguf) | i1-IQ2_M | 29.4 |  |
| [GGUF](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q2_K.gguf) | i1-Q2_K | 31.1 | IQ3_XXS probably better |
| [GGUF](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 31.9 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-IQ3_XS.gguf) | i1-IQ3_XS | 34.0 |  |
| [GGUF](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q3_K_S.gguf) | i1-Q3_K_S | 35.6 | IQ3_XS probably better |
| [GGUF](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q3_K_M.gguf) | i1-Q3_K_M | 39.3 | IQ3_S probably better |
| [GGUF](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q3_K_L.gguf) | i1-Q3_K_L | 42.6 | IQ3_M probably better |
| [GGUF](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-IQ4_XS.gguf) | i1-IQ4_XS | 42.8 |  |
| [GGUF](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q4_0.gguf) | i1-Q4_0 | 45.2 | fast, low quality |
| [GGUF](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q4_K_S.gguf) | i1-Q4_K_S | 45.3 | optimal size/speed/quality |
| [GGUF](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q4_K_M.gguf) | i1-Q4_K_M | 47.8 | fast, recommended |
| [PART 1](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q5_K_S.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q5_K_S.gguf.part2of2) | i1-Q5_K_S | 53.9 |  |
| [PART 1](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q5_K_M.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q5_K_M.gguf.part2of2) | i1-Q5_K_M | 55.4 |  |
| [PART 1](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q6_K.gguf.part1of2) [PART 2](https://huggingface.co/mradermacher/Rhea-72b-v0.5-i1-GGUF/resolve/main/Rhea-72b-v0.5.i1-Q6_K.gguf.part2of2) | i1-Q6_K | 63.4 | practically like static Q6_K |


Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):

![image.png](https://www.nethype.de/huggingface_embed/quantpplgraph.png)

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 -->