base_model: NobodyExistsOnTheInternet/clown-SUV-4x70b
language:
- en
library_name: transformers
license: mit
no_imatrix: >-
[ 5/1523] blk.0.ffn_up.2.weight - [ 8192, 28672, 1,
1], type = f16, converting to iq2_xxs .. Oops: found point 10325 not on
grid: 1 1 1 1 0 2 2 0
quantized_by: mradermacher
About
weighted/imatrix quants of https://huggingface.co/NobodyExistsOnTheInternet/clown-SUV-4x70b
more quants are unlikely to happen because llama.cpp is a crashfest atm. and simply crashes for them
static quants are available at https://huggingface.co/mradermacher/clown-SUV-4x70b-GGUF
Usage
If you are unsure how to use GGUF files, refer to one of TheBloke's READMEs 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 |
|---|---|---|---|
| PART 1 PART 2 | i1-IQ2_M | 78.6 | |
| PART 1 PART 2 | i1-Q2_K | 87.1 | IQ3_XXS probably better |
| PART 1 PART 2 | i1-IQ3_XXS | 91.8 | lower quality |
| PART 1 PART 2 PART 3 | i1-IQ3_XS | 97.1 | |
| PART 1 PART 2 PART 3 | i1-IQ3_S | 103.0 | beats Q3_K* |
| PART 1 PART 2 PART 3 | i1-Q3_K_S | 103.0 | IQ3_XS probably better |
| PART 1 PART 2 PART 3 | i1-IQ3_M | 105.0 | |
| PART 1 PART 2 PART 3 | i1-Q3_K_M | 114.3 | IQ3_S probably better |
| PART 1 PART 2 PART 3 | i1-Q3_K_L | 123.8 | IQ3_M probably better |
| PART 1 PART 2 PART 3 | i1-IQ4_XS | 127.1 | |
| PART 1 PART 2 PART 3 | i1-Q4_0 | 135.0 | fast, low quality |
| PART 1 PART 2 PART 3 | i1-Q4_K_S | 135.7 | optimal size/speed/quality |
| PART 1 PART 2 PART 3 | i1-Q4_K_M | 144.2 | fast, recommended |
| PART 1 PART 2 PART 3 PART 4 | i1-Q5_K_S | 164.1 | |
| PART 1 PART 2 PART 3 PART 4 | i1-Q5_K_M | 169.2 | |
| P1 P2 P3 P4 P5 | i1-Q6_K | 195.7 | 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
FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to questions you might have and/or if you want some other model quantized.
Thanks
I thank my company, nethype GmbH, for letting me use its servers and providing upgrades to my workstation to enable this work in my free time.