Zenabius/calme-3.2-instruct-78b-exl2

EXL2 Quantizations of calme-3.2-instruct-78b

Using exllamav2 release 0.2.6 for quantization.

Original model: https://huggingface.co/MaziyarPanahi/calme-3.2-instruct-78b

Bits 6.5,lm_head 8.0

"quantization_config": {
    "quant_method": "exl2",
    "version": "0.2.6",
    "bits": 6.5,
    "head_bits": 8,
    "calibration": {
        "rows": 115,
        "length": 2048,
        "dataset": "(default)"
    }