Hugging Face's logo

GreenBitAI
/
Phi-3-mini-128k-instruct-layer-mix-bpw-3.0

Text Generation
Transformers
Safetensors
phi3
conversational
custom_code
text-generation-inference
Inference Endpoints
Model card Files Community
Phi-3-mini-128k-instruct-layer-mix-bpw-3.0 / quant_strategy.json
NicoNico6
update
767b93d
raw
history blame contribute delete
41.3 kB
{
"measurement": {
"model.layers.0": {
"accuracy": 0.6637592315673828,
"total_bits": 357441024,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.1": {
"accuracy": 0.7465658187866211,
"total_bits": 357441024,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.2": {
"accuracy": 0.898890495300293,
"total_bits": 357441024,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.3": {
"accuracy": 0.8928320407867432,
"total_bits": 289320576,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.4": {
"accuracy": 0.8153390884399414,
"total_bits": 289320576,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.5": {
"accuracy": 0.9311076402664185,
"total_bits": 289320576,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.6": {
"accuracy": 0.9273414611816406,
"total_bits": 289320576,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.7": {
"accuracy": 0.8751392364501953,
"total_bits": 289320576,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.8": {
"accuracy": 0.872112512588501,
"total_bits": 283125504,
"o_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
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.9": {
"accuracy": 0.8808994293212891,
"total_bits": 233583360,
"o_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
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.10": {
"accuracy": 0.8999626636505127,
"total_bits": 283125504,
"o_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
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.11": {
"accuracy": 0.8984930515289307,
"total_bits": 283125504,
"o_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
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.12": {
"accuracy": 0.8965449333190918,
"total_bits": 252161664,
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.13": {
"accuracy": 0.924396276473999,
"total_bits": 301703808,
"o_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
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.14": {
"accuracy": 0.9207086563110352,
"total_bits": 233583360,
"o_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
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.15": {
"accuracy": 0.9219722747802734,
"total_bits": 233583360,
"o_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
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.16": {
"accuracy": 0.9315023422241211,
"total_bits": 252161664,
"o_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"down_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.17": {
"accuracy": 0.9288318157196045,
"total_bits": 283125504,
"o_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
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.18": {
"accuracy": 0.9211232662200928,
"total_bits": 283125504,
"o_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
},
"qkv_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.19": {
"accuracy": 0.9357051849365234,
"total_bits": 289320576,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.20": {
"accuracy": 0.9271621704101562,
"total_bits": 289320576,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.21": {
"accuracy": 0.9496371746063232,
"total_bits": 357441024,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.22": {
"accuracy": 0.9459362030029297,
"total_bits": 357441024,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"2": 64
},
"bits": [
2
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.23": {
"accuracy": 0.9806646108627319,
"total_bits": 456530304,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.24": {
"accuracy": 0.9800381660461426,
"total_bits": 456530304,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.25": {
"accuracy": 0.980277419090271,
"total_bits": 456530304,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.26": {
"accuracy": 0.9804829359054565,
"total_bits": 456530304,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.27": {
"accuracy": 0.9804953336715698,
"total_bits": 456530304,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.28": {
"accuracy": 0.9789243936538696,
"total_bits": 456530304,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.29": {
"accuracy": 0.9709323644638062,
"total_bits": 456530304,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.30": {
"accuracy": 0.9765185117721558,
"total_bits": 456530304,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
},
"model.layers.31": {
"accuracy": 0.9663764238357544,
"total_bits": 456530304,
"o_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
},
"qkv_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
},
"gate_up_proj": {
"group_size": {
"4": 128
},
"bits": [
4
],
"bits_prop": [
1
],
"scale_bits": 4
}
}
}
}