fix three NPU benchmark issues (#12350)

* fix three issues

* limit mixed_precision for CW only

python/llm/dev/benchmark/all-in-one/run.py

@@ -51,6 +51,8 @@
 
 QWENVL_IDS = ['Qwen/Qwen-VL-Chat']
 
+MINICPM_IDS = ['openbmb/MiniCPM-1B-sft-bf16 ', 'openbmb/MiniCPM-2B-sft-bf16']
+
 MINICPM_V_IDS = ['openbmb/MiniCPM-V-2_6', 'openbmb/MiniCPM-Llama3-V-2_5']
 
 DUMMY_IDS = ['dummy/dummy-1.5B', 'dummy/dummy-4B']
@@ -662,10 +664,11 @@ def transformers_int4_npu_win(repo_id,
             # slice the input_ids to ensure the prompt length is required length.
             input_ids = tokenizer.encode(input_str, return_tensors="pt")
             input_ids = input_ids[:, :in_len]
-            true_str = tokenizer.batch_decode(input_ids)[0]
-            input_list = [true_str] * batch_size
-            input_ids = tokenizer(input_list, return_tensors="pt").input_ids
-            input_ids = input_ids[:, :in_len]
+            if repo_id not in MINICPM_IDS:
+                true_str = tokenizer.batch_decode(input_ids)[0]
+                input_list = [true_str] * batch_size
+                input_ids = tokenizer(input_list, return_tensors="pt").input_ids
+                input_ids = input_ids[:, :in_len]
             actual_in_len = input_ids.shape[1]
             result[in_out] = []
             for i in range(num_trials + warm_up):
@@ -701,14 +704,15 @@ def transformers_int4_npu_pipeline_win(repo_id,
     model_path = get_model_path(repo_id, local_model_hub)
     in_out_len = in_out_pairs[0].split("-")
     max_context_len = max(int(in_out_len[0]) + int(in_out_len[1]), 1024)
+    mixed_precision = True if npu_group_size == 0 else False
     # Load model in 4 bit,
     # which convert the relevant layers in the model into INT4 format
     st = time.perf_counter()
 
     model = AutoModelForCausalLM.from_pretrained(model_path, load_in_low_bit=low_bit, trust_remote_code=True, pipeline=True, torch_dtype=torch.float16,
                                                  optimize_model=optimize_model, max_context_len=max_context_len, max_prompt_len=int(in_out_len[0]), 
                                                  quantization_group_size=npu_group_size, transpose_value_cache=transpose_value_cache,
-                                                 use_cache=True, attn_implementation="eager").eval()
+                                                 use_cache=True, attn_implementation="eager", mixed_precision=mixed_precision).eval()
     tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
 
     end = time.perf_counter()
@@ -726,10 +730,11 @@ def transformers_int4_npu_pipeline_win(repo_id,
             # slice the input_ids to ensure the prompt length is required length.
             input_ids = tokenizer.encode(input_str, return_tensors="pt")
             input_ids = input_ids[:, :in_len]
-            true_str = tokenizer.batch_decode(input_ids)[0]
-            input_list = [true_str] * batch_size
-            input_ids = tokenizer(input_list, return_tensors="pt").input_ids
-            input_ids = input_ids[:, :in_len]
+            if repo_id not in MINICPM_IDS:
+                true_str = tokenizer.batch_decode(input_ids)[0]
+                input_list = [true_str] * batch_size
+                input_ids = tokenizer(input_list, return_tensors="pt").input_ids
+                input_ids = input_ids[:, :in_len]
             actual_in_len = input_ids.shape[1]
             result[in_out] = []
             for i in range(num_trials + warm_up):

python/llm/src/ipex_llm/transformers/npu_pipeline_model/convert_pipeline.py

@@ -118,7 +118,7 @@ def generate(
                                         self.head_dim, self.num_layers,
                                         self.vocab_size,
                                         self.transpose_value_cache,
-                                        new_tokens - 1))
+                                        new_tokens))
         thread.start()
 
         in_pipe_path = "\\\\.\\pipe\\llminputpipe"