Hugging Face's logo

GreenBitAI
/
Llama-2-7B-layer-mix-bpw-2.5

Text Generation
Transformers
Safetensors
llama
text-generation-inference
Inference Endpoints
Model card Files Community
Llama-2-7B-layer-mix-bpw-2.5 / quant_strategy.json
NicoNico6
update
a8a54b0
raw
history blame contribute delete
68.9 kB
{
"measurement": {
"model.layers.0": {
"accuracy": 0.9122365315755208,
"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.1": {
"accuracy": 0.9080530007680258,
"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.2": {
"accuracy": 0.9275489846865336,
"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.3": {
"accuracy": 0.8821520010630289,
"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": 0.8701377709706625,
"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.5": {
"accuracy": 0.857724110285441,
"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": 0.8618481953938801,
"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.7": {
"accuracy": 0.8569034735361736,
"total_bits": 483482688,
"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": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.8": {
"accuracy": 0.8387942314147949,
"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.9": {
"accuracy": 0.8487000465393066,
"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.10": {
"accuracy": 0.8522807757059734,
"total_bits": 483482688,
"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": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.11": {
"accuracy": 0.8486220041910807,
"total_bits": 483482688,
"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": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.12": {
"accuracy": 0.8406097094217937,
"total_bits": 483482688,
"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": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.13": {
"accuracy": 0.8556122779846191,
"total_bits": 483482688,
"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": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.14": {
"accuracy": 0.8522577285766602,
"total_bits": 483482688,
"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": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.15": {
"accuracy": 0.8361595471700033,
"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.16": {
"accuracy": 0.8548601468404134,
"total_bits": 483482688,
"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": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.17": {
"accuracy": 0.8503810564676921,
"total_bits": 483482688,
"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": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.18": {
"accuracy": 0.8818686803181967,
"total_bits": 539218464,
"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": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.19": {
"accuracy": 0.8884480794270834,
"total_bits": 539218464,
"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": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.20": {
"accuracy": 0.8379961649576823,
"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.21": {
"accuracy": 0.8830145200093588,
"total_bits": 539218464,
"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": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.22": {
"accuracy": 0.8894084294637045,
"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.23": {
"accuracy": 0.870563824971517,
"total_bits": 539218464,
"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": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.24": {
"accuracy": 0.8652655283610026,
"total_bits": 539218464,
"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": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.25": {
"accuracy": 0.8744535446166992,
"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.26": {
"accuracy": 0.8678053220113119,
"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.27": {
"accuracy": 0.8675735791524252,
"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.28": {
"accuracy": 0.860187848409017,
"total_bits": 539218464,
"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": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.29": {
"accuracy": 0.8641198476155599,
"total_bits": 539218464,
"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": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.30": {
"accuracy": 0.9148619969685872,
"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.31": {
"accuracy": 0.8672019640604655,
"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
}
}
}
}