For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Python class

GemmaRMSNorm

GemmaRMSNorm

class max.experimental.nn.norm.GemmaRMSNorm(dim, eps=1e-06)

source

Bases: RMSNorm

Gemma-style root mean square normalization.

Subclasses RMSNorm with two differences:

• Scales by 1 + weight rather than weight.
• Multiplies by the scale before casting back to the input dtype, instead of after.

The constructor signature is identical to RMSNorm. Used by the Gemma model family.

Parameters:

• dim (int)
• eps (float)

forward()

forward(x)

source

Returns x normalized using the Gemma-style RMS variant.

Parameters:

x (Tensor)

Return type:

Tensor

Was this page helpful?

Thank you! We'll create more content like this.

Thank you for helping us improve!

😔 What went wrong?

Some code doesn’t work

It includes inaccurate information

It's missing information I need

It was difficult to understand

Other

Submit