Hugging Face's logo

GreenBitAI
/
Llama-2-7B-layer-mix-bpw-3.0

Text Generation
Transformers
Safetensors
llama
text-generation-inference
Inference Endpoints
Model card Files Community
Llama-2-7B-layer-mix-bpw-3.0 / quant_strategy.json
NicoNico6
update
f3de630
raw
history blame contribute delete
68.9 kB
{
"measurement": {
"model.layers.0": {
"accuracy": 0.9575417041778564,
"total_bits": 638304288,
"q_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.1": {
"accuracy": 0.9234476089477539,
"total_bits": 694042656,
"q_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.2": {
"accuracy": 0.9642077088356018,
"total_bits": 417425472,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.3": {
"accuracy": 0.9589163064956665,
"total_bits": 450454080,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.4": {
"accuracy": 0.9506402015686035,
"total_bits": 417425472,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.5": {
"accuracy": 0.9490039348602295,
"total_bits": 450454080,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.6": {
"accuracy": 0.9650095701217651,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.7": {
"accuracy": 0.961586594581604,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.8": {
"accuracy": 0.96183180809021,
"total_bits": 605275680,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.9": {
"accuracy": 0.9578591585159302,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.10": {
"accuracy": 0.955805778503418,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.11": {
"accuracy": 0.9599390029907227,
"total_bits": 605275680,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.12": {
"accuracy": 0.9548064470291138,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.13": {
"accuracy": 0.954245924949646,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.14": {
"accuracy": 0.9533202648162842,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.15": {
"accuracy": 0.9508509635925293,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.16": {
"accuracy": 0.9490735530853271,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.17": {
"accuracy": 0.950542688369751,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.18": {
"accuracy": 0.947782039642334,
"total_bits": 572247072,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.19": {
"accuracy": 0.9630944728851318,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.20": {
"accuracy": 0.9612696170806885,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.21": {
"accuracy": 0.9627827405929565,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.22": {
"accuracy": 0.9584479331970215,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.23": {
"accuracy": 0.9615379571914673,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.24": {
"accuracy": 0.960837721824646,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.25": {
"accuracy": 0.9591341018676758,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.26": {
"accuracy": 0.957118034362793,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.27": {
"accuracy": 0.9584870338439941,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.28": {
"accuracy": 0.9577126502990723,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.29": {
"accuracy": 0.9580142498016357,
"total_bits": 661014048,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.30": {
"accuracy": 0.9636353254318237,
"total_bits": 749781024,
"q_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.31": {
"accuracy": 0.9668803215026855,
"total_bits": 815838240,
"q_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"k_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"v_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
}
}
}