Add phi3RMS (#10988)

* phi3RMS

python/llm/src/ipex_llm/transformers/convert.py

@@ -1517,6 +1517,13 @@ def _optimize_post(model, lightweight_bmm=False):
         from ipex_llm.transformers.models.phi3 import model_forward_wrapper
         model_forward = model_forward_wrapper(module.Phi3Model.forward)
         convert_forward(model, module.Phi3Model, model_forward)
+        from ipex_llm.transformers.models.phi3 import phi3_rms_norm_forward
+        convert_forward(
+            model,
+            module.Phi3RMSNorm,
+            phi3_rms_norm_forward)
+        # Empty cache after the first attention to run long context.
+        model.model.layers[0].self_attn.register_forward_hook(empty_cache_post)
     elif model.config.model_type == 'yuan':
         modeling_module_name = model.__class__.__module__
         module = importlib.import_module(modeling_module_name)

python/llm/src/ipex_llm/transformers/models/phi3.py

@@ -253,3 +253,17 @@ def model_forward_wrapper(origin_model_forward):
             return_dict=return_dict,
         )
     return model_forward
+
+
+def phi3_rms_norm_forward(self, hidden_states):
+    if hidden_states.device.type == "xpu" and not (self.training and hidden_states.requires_grad):
+        import linear_q4_0
+        x_2d = hidden_states.reshape(-1, hidden_states.size(-1)).contiguous()
+        output = linear_q4_0.rms_norm(self.weight, x_2d, self.variance_epsilon)
+        return output.reshape(hidden_states.shape)
+
+    input_dtype = hidden_states.dtype
+    hidden_states = hidden_states.to(torch.float32)
+    variance = hidden_states.pow(2).mean(-1, keepdim=True)
+    hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon)
+    return self.weight * hidden_states.to(input_dtype)