add rms eps to init_weights

modeling_glm2.py

@@ -373,6 +373,10 @@ class gLM2PreTrainedModel(PreTrainedModel):
             # Force the buffer to update
             with torch.no_grad():
                 module.inv_freq.copy_(inv_freq)
+        elif isinstance(module, RMSNorm):
+            if hasattr(module, "variance_epsilon"):
+                with torch.no_grad():
+                    module.variance_epsilon.fill_(self.config.norm_eps)
 
 
 class gLM2Model(gLM2PreTrainedModel):
@@ -476,4 +480,4 @@ class gLM2LMHead(nn.Module):
             config.dim, config.vocab_size, bias=False)
 
     def forward(self, features):
-        return self.proj_output(self.norm(features))
+        return self.proj_output(self.norm(features))