MCPcopy Index your code
hub / github.com/deepspeedai/DeepSpeedExamples / __init__

Method __init__

bing_bert/nvidia/modeling.py:296–302  ·  view source on GitHub ↗

Construct a layernorm module in the TF style (epsilon inside the square root).

(self, hidden_size, eps=1e-12)

Source from the content-addressed store, hash-verified

294 #print("Better speed can be achieved with apex installed from https://www.github.com/nvidia/apex.")
295 class BertLayerNorm(nn.Module):
296 def __init__(self, hidden_size, eps=1e-12):
297 """Construct a layernorm module in the TF style (epsilon inside the square root).
298 """
299 super(BertLayerNorm, self).__init__()
300 self.weight = nn.Parameter(torch.ones(hidden_size))
301 self.bias = nn.Parameter(torch.zeros(hidden_size))
302 self.variance_epsilon = eps
303
304 def forward(self, x):
305 u = x.mean(-1, keepdim=True)

Callers

nothing calls this directly

Calls 1

__init__Method · 0.45

Tested by

no test coverage detected