bartowski/dolphin-2.6-mistral-7b-dpo-exl2

Exllama v2 Quantizations of dolphin-2.6-mistral-7b-dpo

Using turboderp's ExLlamaV2 v0.0.11 for quantization.

Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.

Original model: https://huggingface.co/cognitivecomputations/dolphin-2.6-mistral-7b-dpo

Model Size: 7b

Branch	Bits	lm_head bits	Dataset	Size	Description
8_0	8.0	8.0	Default	9.8 GB	Maximum quality that ExLlamaV2 can produce, near unquantized performance.
6_5	6.5	8.0	Default	8.6 GB	Very similar to 8.0, good tradeoff of size vs performance, recommended.
5_0	5.0	6.0	Default	7.4 GB	Slightly lower perplexity vs 6.5.
4_0	4.0	6.0	Default	6.5 GB	Just under GPTQ equivalent bits per weight.

All VRAM requirements estimated from 16k context. For 32k context add ~2 GB.

4.0 bits per weight

5.0 bits per weight

6.5 bits per weight

8.0 bits per weight

Download instructions

With git:

git clone --single-branch --branch 4_0 https://huggingface.co/bartowski/dolphin-2.6-mistral-7b-dpo-exl2

With huggingface hub (credit to TheBloke for instructions):

pip3 install huggingface-hub

To download the main (only useful if you only care about measurement.json) branch to a folder called dolphin-2.6-mistral-7b-dpo-exl2:

mkdir dolphin-2.6-mistral-7b-dpo-exl2
huggingface-cli download bartowski/dolphin-2.6-mistral-7b-dpo-exl2 --local-dir dolphin-2.6-mistral-7b-dpo-exl2 --local-dir-use-symlinks False

To download from a different branch, add the --revision parameter:

mkdir dolphin-2.6-mistral-7b-dpo-exl2-6_5
huggingface-cli download bartowski/dolphin-2.6-mistral-7b-dpo-exl2 --revision 6_5 --local-dir dolphin-2.6-mistral-7b-dpo-exl2-6_5 --local-dir-use-symlinks False