optimize Minicpm (intel-analytics#11189)

* minicpm optimize * update
SichengStevenLi · Jun 3, 2024 · 5f13700 · 5f13700
1 parent ff83fad
commit 5f13700
Show file tree

Hide file tree

Showing 2 changed files with 380 additions and 0 deletions.
diff --git a/python/llm/src/ipex_llm/transformers/convert.py b/python/llm/src/ipex_llm/transformers/convert.py
@@ -1598,4 +1598,18 @@ def safe_bmm_fwd(*args, **kwargs):
                         module.StableLmModel,
                         stablelm_model_forward
                         )
+    elif model.config.model_type == 'minicpm':
+        from ipex_llm.transformers.models.minicpm import minicpm_attention_forward
+        modeling_module_name = model.__class__.__module__
+        module = importlib.import_module(modeling_module_name)
+        convert_forward(model,
+                        module.MiniCPMMLP,
+                        llama_mlp_forward)
+        convert_forward(model,
+                        module.MiniCPMRMSNorm,
+                        llama_rms_norm_forward)
+        convert_forward(model,
+                        module.MiniCPMAttention,
+                        minicpm_attention_forward)
+
     return model