ipt-350m / quantize_config.json
Xenova's picture
Xenova HF staff
Upload folder using huggingface_hub
20056e4
{
"per_channel": false,
"reduce_range": false,
"per_model_config": {
"decoder_model": {
"op_types": [
"Cast",
"Sub",
"Div",
"ReduceMean",
"Pow",
"Erf",
"Gather",
"Less",
"Add",
"Where",
"Equal",
"Mul",
"Or",
"Slice",
"Range",
"ConstantOfShape",
"Transpose",
"MatMul",
"Softmax",
"Not",
"Constant",
"ScatterND",
"Expand",
"Reshape",
"Sqrt",
"Unsqueeze",
"Concat",
"Shape"
],
"weight_type": "QInt8"
},
"decoder_model_merged": {
"op_types": [
"Cast",
"Sub",
"Div",
"ReduceMean",
"Pow",
"Erf",
"Gather",
"Less",
"Add",
"Where",
"Equal",
"Mul",
"Slice",
"Or",
"Range",
"ConstantOfShape",
"Transpose",
"MatMul",
"Softmax",
"Not",
"Constant",
"ScatterND",
"Expand",
"Reshape",
"If",
"Sqrt",
"Unsqueeze",
"Concat",
"Shape"
],
"weight_type": "QInt8"
},
"decoder_with_past_model": {
"op_types": [
"Div",
"Cast",
"Sub",
"Pow",
"ReduceMean",
"Erf",
"Gather",
"Add",
"Where",
"Equal",
"Mul",
"Slice",
"ConstantOfShape",
"Transpose",
"MatMul",
"Softmax",
"Not",
"Constant",
"Expand",
"Reshape",
"Sqrt",
"Unsqueeze",
"Concat",
"Shape"
],
"weight_type": "QInt8"
}
}
}