--- license: llama3.3 base_model: - rombodawg/Rombos-LLM-70b-Llama-3.3 base_model_relation: quantized library_name: transformers quantized_by: bartowski pipeline_tag: text-generation --- ## Exllama v2 Quantizations of Rombos-LLM-70b-Llama-3.3 Using turboderp's ExLlamaV2 v0.2.4 for quantization. The "main" branch only contains the measurement.json, download one of the other branches for the model (see below) Each branch contains an individual bits per weight, with the main one containing only the meaurement.json for further conversions. Conversion was done using the default 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/rombodawg/Rombos-LLM-70b-Llama-3.3 8.0 bits per weight 6.5 bits per weight 4.25 bits per weight 3.5 bits per weight 3.0 bits per weight 2.2 bits per weight ## Download instructions With git: ```shell git clone --single-branch --branch 6_5 https://huggingface.co/bartowski/Rombos-LLM-70b-Llama-3.3-exl2 ``` With huggingface hub (credit to TheBloke for instructions): ```shell pip3 install huggingface-hub ``` To download the `main` (only useful if you only care about measurement.json) branch to a folder called `Rombos-LLM-70b-Llama-3.3-exl2`: ```shell mkdir Rombos-LLM-70b-Llama-3.3-exl2 huggingface-cli download bartowski/Rombos-LLM-70b-Llama-3.3-exl2 --local-dir Rombos-LLM-70b-Llama-3.3-exl2 ``` To download from a different branch, add the `--revision` parameter: Linux: ```shell mkdir Rombos-LLM-70b-Llama-3.3-exl2-6_5 huggingface-cli download bartowski/Rombos-LLM-70b-Llama-3.3-exl2 --revision 6_5 --local-dir Rombos-LLM-70b-Llama-3.3-exl2-6_5 ``` Windows (which apparently doesn't like _ in folders sometimes?): ```shell mkdir Rombos-LLM-70b-Llama-3.3-exl2-6.5 huggingface-cli download bartowski/Rombos-LLM-70b-Llama-3.3-exl2 --revision 6_5 --local-dir Rombos-LLM-70b-Llama-3.3-exl2-6.5 ```