pardonmyai-tiny / quantize_config.json
tarekziade's picture
Upload 9 files
eae722d verified
raw
history blame
722 Bytes
{
"per_channel": true,
"reduce_range": true,
"per_model_config": {
"model": {
"op_types": [
"Concat",
"Tanh",
"Add",
"MatMul",
"Mul",
"Erf",
"Gemm",
"Softmax",
"Shape",
"Transpose",
"ReduceMean",
"Sub",
"Reshape",
"Cast",
"Pow",
"Div",
"Gather",
"Slice",
"Constant",
"Unsqueeze",
"Sqrt"
],
"weight_type": "QInt8"
}
}
}