base_model: deepseek-ai/deepseek-llm-67b-chat
language:
- en
library_name: transformers
license: other
license_link: LICENSE
license_name: deepseek
quantized_by: mradermacher
About
weighted/imatrix quants of https://huggingface.co/deepseek-ai/deepseek-llm-67b-chat
static quants are available at https://huggingface.co/mradermacher/deepseek-llm-67b-chat-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 |
---|---|---|---|
GGUF | i1-IQ1_S | 14.8 | for the desperate |
GGUF | i1-IQ1_M | 16.1 | mostly desperate |
GGUF | i1-IQ2_XXS | 18.3 | |
GGUF | i1-IQ2_XS | 20.3 | |
GGUF | i1-IQ2_S | 21.4 | |
GGUF | i1-IQ2_M | 23.2 | |
GGUF | i1-Q2_K | 25.2 | IQ3_XXS probably better |
GGUF | i1-IQ3_XXS | 26.2 | lower quality |
GGUF | i1-IQ3_XS | 28.0 | |
GGUF | i1-Q3_K_S | 29.4 | IQ3_XS probably better |
GGUF | i1-IQ3_S | 29.5 | beats Q3_K* |
GGUF | i1-IQ3_M | 30.6 | |
GGUF | i1-Q3_K_M | 32.8 | IQ3_S probably better |
GGUF | i1-Q3_K_L | 35.7 | IQ3_M probably better |
GGUF | i1-IQ4_XS | 36.3 | |
GGUF | i1-Q4_0 | 38.4 | fast, low quality |
GGUF | i1-Q4_K_S | 38.5 | optimal size/speed/quality |
GGUF | i1-Q4_K_M | 40.5 | fast, recommended |
GGUF | i1-Q5_K_S | 46.6 | |
GGUF | i1-Q5_K_M | 47.8 | |
PART 1 PART 2 | i1-Q6_K | 55.4 | 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.