Exllama v2 Quantizations of mistral-7b-v0.2-layla-v4
Using turboderp's ExLlamaV2 v0.0.16 for quantization.
Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions.
Original model: l3utterfly/mistral-7b-v0.2-layla-v4
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.
Download instructions
With git:
git clone --single-branch --branch 4_0 https://huggingface.co/suparious/mistral-7b-v0.2-layla-v4-exl2
With huggingface hub (credit to TheBloke and bartowski for instructions):
pip3 install huggingface-hub
To download the main (only useful if you only care about measurement.json) branch to a folder called mistral-7b-v0.2-layla-v4-exl2:
mkdir mistral-7b-v0.2-layla-v4-exl2
huggingface-cli download suparious/mistral-7b-v0.2-layla-v4-exl2 --local-dir mistral-7b-v0.2-layla-v4-exl2 --local-dir-use-symlinks False
To download from a different branch, add the --revision parameter:
mkdir mistral-7b-v0.2-layla-v4-exl2-6_5
huggingface-cli download suparious/mistral-7b-v0.2-layla-v4-exl2 --revision 6_5 --local-dir mistral-7b-v0.2-layla-v4-exl2-6_5 --local-dir-use-symlinks False
Inference Providers
NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API:
The model has no library tag.