--is_weight_clip

Applicability

Product	Supported
Atlas A3 training products/Atlas A3 inference products	√
Atlas A2 training products/Atlas A2 inference products	√
Atlas 200I/500 A2 inference products	√
Atlas inference products	√
Atlas training products	√

Description

Clips data when the upper bits of the floating-point weight data are converted to lower bits.

If the weight data type of the original model is high (for example, float32), the cast operator is inserted during model conversion to convert the data type from high (for example, float32) to low (for example, float16). In this scenario, data overflow may occur, by enabling the --is_weight_clip option, the upper-bit (for example, float32) data is cropped before the cast operator to prevent data overflow.

Argument

0: disabled
1 (default): enabled

Suggestions and Benefits

None

Example

--is_weight_clip=1

Dependencies and Restrictions