[NPU] Support minicpm-v with python cpp backend (#12637)

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

@@ -650,7 +650,7 @@ def transformers_int4_npu_win(repo_id,
     load_time = end - st
     print(">> loading of model costs {}s".format(load_time))
 
-    if not hasattr(model, "model_ptr"):
+    if not hasattr(model, "model_ptr") or repo_id in MINICPM_V_IDS:
         model = BenchmarkWrapper(model)
 
     result = {}

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

@@ -301,8 +301,7 @@ class _BaseAutoModelClass:
         model.share_memory()
 
         if not pipeline:
-            if (not hasattr(model, 'llm') and
+            if model.config.model_type in ["qwen2", "llama", "minicpm"]:
-                    model.config.model_type in ["qwen2", "llama", "minicpm"]):
                 from ipex_llm.transformers.npu_models.convert import optimize_llm_single_process
                 optimize_llm_single_process(
                     llm,
@@ -312,7 +311,8 @@ class _BaseAutoModelClass:
                     group_size=quantization_group_size,
                     qtype=qtype,
                     save_directory=save_directory,
-                    fuse_layers=fuse_layers
+                    fuse_layers=fuse_layers,
+                    has_llm=hasattr(model, "llm")
                 )
             else:
                 optimize_llm(

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

@@ -449,7 +449,8 @@ def optimize_llm_single_process(
     group_size: int,
     qtype: str,
     save_directory: str,
-    fuse_layers: int=None
+    fuse_layers: int=None,
+    has_llm: bool=False
 ):
     from ipex_llm.transformers.npu_pipeline_model.convert_pipeline import convert_llm
     from .npu_llm_cpp import load_model_from_file
@@ -468,8 +469,13 @@ def optimize_llm_single_process(
         model.kv_len = kv_len
         model.model_ptr = model_ptr
         model.save_directory = save_directory
+        if model.config.vocab_size == 151666:
+            # for MiniCPM-V 2.6, 152064 is vocab_size of Qwen2-7B
+            model.vocab_size = 152064
+        else:
             model.vocab_size = model.config.vocab_size
         model.logits_buffer = torch.empty(1, 1, model.vocab_size, dtype=torch.float32)
+        model.max_prompt_len = max_prompt_len
     except:
         invalidInputError(False,
                           "False to InitLLMPipeline.")
@@ -478,6 +484,7 @@ def optimize_llm_single_process(
     general_convert(model, PreTrainedModel, prepare_input_ids, "prepare_inputs_for_generation")
     general_convert(model, PreTrainedModel, causal_lm_forward)
     # patch generate function
+    if not has_llm:
         import types
         model.original_generate = model.generate
         model.generate = types.MethodType(generate, model)
@@ -491,9 +498,10 @@ def prepare_input_ids(
     else:  # prefill, reset the model here
         from .npu_llm_cpp import reset
         reset(self.model_ptr)
-    model_inputs = {
+    if inputs_embeds is not None and past_key_values is None:
-        "input_ids": input_ids
+        model_inputs = {"inputs_embeds": inputs_embeds}
-    }
+    else:
+        model_inputs = {"input_ids": input_ids}
     return model_inputs
 
 
@@ -511,6 +519,7 @@ def causal_lm_forward(
     return_dict: Optional[bool] = None,
 ) -> Union[Tuple, CausalLMOutputWithPast]:
     from .npu_llm_cpp import run_prefill_with_logits, run_decode_with_logits
+    if input_ids is not None:
         if isinstance(input_ids[0], torch.Tensor):
             input_list = input_ids[0].flatten().tolist()
         else:
@@ -522,6 +531,19 @@ def causal_lm_forward(
         else:
             logits = run_decode_with_logits(self.model_ptr, input_list[0],
                                             self.logits_buffer, self.vocab_size)
+    elif inputs_embeds is not None:
+        seq_len = inputs_embeds.shape[1]
+        pad_len = self.max_prompt_len - seq_len
+        inputs_embeds = torch.nn.functional.pad(inputs_embeds.to(torch.float16),
+                                                (0, 0, 0, pad_len), value=0.0)
+        logits = run_prefill_with_logits(self.model_ptr, None, self.logits_buffer,
+                                         self.vocab_size, inputs_embeds, seq_len)
+    else:
+        invalidInputError(False, "Please specify either input_ids or inputs_embeds.")
+
+    if self.config.vocab_size == 151666:
+        # for MiniCPM-V 2.6
+        logits = logits[:, :, :151666]
 
     return CausalLMOutputWithPast(
         loss=None,

python/llm/src/ipex_llm/transformers/npu_models/npu_llm_cpp.py

@@ -48,8 +48,8 @@ _lib = ctypes.cdll.LoadLibrary(_lib_path)
 _lib.load_model_from_file.argtypes = [ctypes.c_char_p]
 _lib.load_model_from_file.restype = ctypes.c_void_p
 
-_lib.run_prefill.argtypes = [ctypes.c_void_p, ctypes.POINTER(ctypes.c_int), ctypes.c_int,
+_lib.run_prefill.argtypes = [ctypes.c_void_p, ctypes.c_void_p, ctypes.c_int,
-                             ctypes.c_float]
+                             ctypes.c_float, ctypes.c_bool]
 _lib.run_prefill.restype = ctypes.POINTER(ctypes.c_float)
 
 _lib.run_decode.argtypes = [ctypes.c_void_p, ctypes.c_int, ctypes.c_float]
@@ -61,8 +61,10 @@ _lib.llm_sample_token.restype = ctypes.c_int
 _lib.reset.argtypes = [ctypes.c_void_p]
 _lib.reset.restype = None
 
-_lib.run_prefill_with_logits.argtypes = [ctypes.c_void_p, ctypes.POINTER(ctypes.c_int),
+_lib.run_prefill_with_logits.argtypes = [ctypes.c_void_p, ctypes.c_void_p,
-                                         ctypes.c_int, ctypes.POINTER(ctypes.c_float), ctypes.c_int]
+                                         ctypes.c_int, ctypes.POINTER(ctypes.c_float),
+                                         ctypes.c_int, ctypes.c_bool]
+
 _lib.run_prefill_with_logits.restype = None
 
 _lib.run_decode_with_logits.argtypes = [ctypes.c_void_p, ctypes.c_int,
@@ -77,7 +79,7 @@ def load_model_from_file(model_dir: str):
 def run_prefill(model_ptr, input_ids, vocab_size, repetition_penalty=1.0):
     input_ptr = (ctypes.c_int32 * len(input_ids))(*input_ids)
     input_len = len(input_ids)
-    plogits = _lib.run_prefill(model_ptr, input_ptr, input_len, repetition_penalty)
+    plogits = _lib.run_prefill(model_ptr, input_ptr, input_len, repetition_penalty, False)
     new_token = _lib.llm_sample_token(plogits, True, vocab_size)
     return new_token
 
@@ -88,12 +90,19 @@ def run_decode(model_ptr, input_id, vocab_size, repetition_penalty=1.0):
     return new_token
 
 
-def run_prefill_with_logits(model_ptr, input_ids, logits, vocab_size):
+def run_prefill_with_logits(model_ptr, input_ids, logits, vocab_size,
+                            inputs_embeds=None, seq_len=None):
+    if input_ids is not None:
         input_ptr = (ctypes.c_int32 * len(input_ids))(*input_ids)
         input_len = len(input_ids)
+    else:
+        input_ptr = inputs_embeds.contiguous().data.data_ptr()
+        input_ptr = ctypes.cast(input_ptr, ctypes.c_void_p)
+        input_len = seq_len
     logits_ptr = logits.data.data_ptr()
     logits_ptr = ctypes.cast(logits_ptr, ctypes.POINTER(ctypes.c_float))
-    _lib.run_prefill_with_logits(model_ptr, input_ptr, input_len, logits_ptr, vocab_size)
+    _lib.run_prefill_with_logits(model_ptr, input_ptr, input_len, logits_ptr,
+                                 vocab_size, (input_ids is None))
     return logits
 
 

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

@@ -34,6 +34,10 @@ def convert_lm_head_and_embedding(model, temp_dir, weight_dir,
     lm_head_n_splits = 1
     asym = getattr(model.config, "asym", False)
 
+    if vocab_size == 151666:
+        # for MiniCPM-V 2.6 lm_head on NPU
+        vocab_size = 152064
+
     if not isinstance(lm_head, SlicedLMHead):
         asym = lm_head.qtype == "asym_int4_rtn"
         if asym: