nlp-qual-qual / quantize_config.json
kiranbandi's picture
Upload 8 files
ca51d21 verified
{
"per_channel": true,
"reduce_range": true,
"per_model_config": {
"model": {
"op_types": [
"Erf",
"Concat",
"Sub",
"Unsqueeze",
"Constant",
"Gemm",
"ReduceMean",
"Reshape",
"Transpose",
"MatMul",
"Div",
"Slice",
"Mul",
"Tanh",
"Pow",
"Cast",
"Shape",
"Gather",
"Add",
"Softmax",
"Sqrt"
],
"weight_type": "QInt8"
}
}
}