TinyLlama-1.1B-Chat-v0.5-exl2

Exllama v2 Quantizations of TinyLlama-1.1B-Chat-v0.5

Using <a href="https://github.com/turboderp/exllamav2/releases/tag/v0.0.8">turboderp's ExLlamaV2 v0.0.8</a> for quantization.

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

Conversion was done using wikitext-103-raw-v1-test.parquet as calibration dataset.

Default arguments used except when the bits per weight is above 6.0, at that point the lm_head layer is quantized at 8 bits per weight instead of the default 6.

Original model: https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v0.5

<a href="https://huggingface.co/bartowski/TinyLlama-1.1B-Chat-v0.5-exl2/tree/4_0">4.0 bits per weight</a>

<a href="https://huggingface.co/bartowski/TinyLlama-1.1B-Chat-v0.5-exl2/tree/6_0">6.0 bits per weight</a>

<a href="https://huggingface.co/bartowski/TinyLlama-1.1B-Chat-v0.5-exl2/tree/8_0">8.0 bits per weight</a>

Download instructions

With git:

git clone --single-branch --branch 4_0 https://huggingface.co/bartowski/TinyLlama-1.1B-Chat-v0.5-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 TinyLlama-1.1B-Chat-v0.5-exl2:

mkdir TinyLlama-1.1B-Chat-v0.5-exl2
huggingface-cli download bartowski/TinyLlama-1.1B-Chat-v0.5-exl2 --local-dir TinyLlama-1.1B-Chat-v0.5-exl2 --local-dir-use-symlinks False

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

mkdir TinyLlama-1.1B-Chat-v0.5-exl2
huggingface-cli download bartowski/TinyLlama-1.1B-Chat-v0.5-exl2 --revision 4_0 --local-dir TinyLlama-1.1B-Chat-v0.5-exl2 --local-dir-use-symlinks False