<|begin_of_text|><|start_header_id|>system<|end_header_id|>
{system_prompt}<|end_of_text|><|start_header_id|>user<|end_header_id|>
{prompt}<|end_of_text|><|start_header_id|>assistant<|end_header_id|>
| Branch | Bits | lm_head bits | VRAM (4k) | VRAM (8K) | VRAM (16k) | VRAM (32k) | Description |
|---|---|---|---|---|---|---|---|
| 8_0 | 8.0 | 8.0 | 10.1 GB | 10.5 GB | 11.5 GB | 13.6 GB | Maximum quality that ExLlamaV2 can produce, near unquantized performance. |
| 6_5 | 6.5 | 8.0 | 8.9 GB | 9.3 GB | 10.3 GB | 12.4 GB | Very similar to 8.0, good tradeoff of size vs performance, recommended. |
| 5_0 | 5.0 | 6.0 | 7.7 GB | 8.1 GB | 9.1 GB | 11.2 GB | Slightly lower quality vs 6.5, but usable on 8GB cards. |
| 4_25 | 4.25 | 6.0 | 7.0 GB | 7.4 GB | 8.4 GB | 10.5 GB | GPTQ equivalent bits per weight, slightly higher quality. |
| 3_5 | 3.5 | 6.0 | 6.4 GB | 6.8 GB | 7.8 GB | 9.9 GB | Lower quality, only use if you have to. |
git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/Llama-3-8B-LexiFun-Uncensored-V1-exl2 Llama-3-8B-LexiFun-Uncensored-V1-exl2-6_5pip3 install huggingface-hub--revision parameter. For example, to download the 6.5 bpw branch:huggingface-cli download bartowski/Llama-3-8B-LexiFun-Uncensored-V1-exl2 --revision 6_5 --local-dir Llama-3-8B-LexiFun-Uncensored-V1-exl2-6_5 --local-dir-use-symlinks Falsehuggingface-cli download bartowski/Llama-3-8B-LexiFun-Uncensored-V1-exl2 --revision 6_5 --local-dir Llama-3-8B-LexiFun-Uncensored-V1-exl2-6.5 --local-dir-use-symlinks False