Update benchmark script for NPU

python/llm/dev/benchmark/all-in-one/config.yaml

@@ -11,6 +11,7 @@ low_bit: 'sym_int4' # default to use 'sym_int4' (i.e. symmetric int4)
 batch_size: 1 # default to 1
 in_out_pairs:
   - '32-32'
+  - '960-64'
   - '1024-128'
 test_api:
   - "transformer_int4_fp16_gpu"             # on Intel GPU, transformer-like API, (qtype=int4), (dtype=fp16)
@@ -37,5 +38,6 @@ test_api:
   # - "transformers_int4_npu_win"           # on Intel NPU for Windows,  transformer-like API, (qtype=int4)
 cpu_embedding: False # whether put embedding to CPU
 streaming: False # whether output in streaming way (only available now for gpu win related test_api)
+optimize_model: False # whether apply further optimization on NPU (only available now for transformers_int4_npu_win test_api)
 use_fp16_torch_dtype: True # whether use fp16 for non-linear layer (only available now for "pipeline_parallel_gpu" test_api)
 task: 'continuation' # task can be 'continuation', 'QA' and 'summarize'

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

@@ -136,7 +136,7 @@ def preprocess_prompt(tokenizer, in_len, task):
         input_ids = tokenizer.encode(input_str, return_tensors="pt")    
     return input_ids
 
-def run_model(repo_id, test_api, in_out_pairs, local_model_hub=None, warm_up=1, num_trials=3, num_beams=1, low_bit='sym_int4', cpu_embedding=False, batch_size=1, streaming=False, use_fp16_torch_dtype=False, lookahead=False, task='continuation'):
+def run_model(repo_id, test_api, in_out_pairs, local_model_hub=None, warm_up=1, num_trials=3, num_beams=1, low_bit='sym_int4', cpu_embedding=False, batch_size=1, streaming=False, use_fp16_torch_dtype=False, lookahead=False, task='continuation', optimize_model=False):
     # TODO: make a parameter
     result= {}
     if test_api == 'transformer_int4':
@@ -188,7 +188,7 @@ def run_model(repo_id, test_api, in_out_pairs, local_model_hub=None, warm_up=1,
     elif test_api == 'pipeline_parallel_gpu':
         result = run_pipeline_parallel_gpu(repo_id, local_model_hub, in_out_pairs, warm_up, num_trials, num_beams, low_bit, batch_size, cpu_embedding, fp16=use_fp16_torch_dtype)
     elif test_api == 'transformers_int4_npu_win':
-        result = transformers_int4_npu_win(repo_id, local_model_hub, in_out_pairs, warm_up, num_trials, num_beams, low_bit, batch_size)
+        result = transformers_int4_npu_win(repo_id, local_model_hub, in_out_pairs, warm_up, num_trials, num_beams, low_bit, batch_size, optimize_model)
     else:
         invalidInputError(False, "Unknown test_api " + test_api + ", please check your config.yaml.")
 
@@ -603,24 +603,30 @@ def transformers_int4_npu_win(repo_id,
                                  num_trials,
                                  num_beams,
                                  low_bit,
-                                 batch_size):
+                                 batch_size,
+                                 optimize_model):
     from ipex_llm.transformers.npu_model import AutoModel, AutoModelForCausalLM
     from transformers import AutoTokenizer, LlamaTokenizer
 
     model_path = get_model_path(repo_id, local_model_hub)
+    in_out_len = in_out_pairs[0].split("-")
+    max_output_len = max(int(in_out_len[0]) + int(in_out_len[1]), 1024)
     # Load model in 4 bit,
     # which convert the relevant layers in the model into INT4 format
     st = time.perf_counter()
     if repo_id in CHATGLM_IDS:
         model = AutoModel.from_pretrained(model_path, load_in_low_bit=low_bit, trust_remote_code=True,
+                                          optimize_model=optimize_model, max_output_len=max_output_len, max_prompt_len=int(in_out_len[0]), transpose_value_cache=True,
                                           torch_dtype='auto', attn_implementation="eager").eval()
         tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
     elif repo_id in LLAMA_IDS:
         model = AutoModelForCausalLM.from_pretrained(model_path, load_in_low_bit=low_bit, trust_remote_code=True,
+                                                     optimize_model=optimize_model, max_output_len=max_output_len, max_prompt_len=int(in_out_len[0]), transpose_value_cache=True,
                                                      use_cache=True, attn_implementation="eager").eval()
         tokenizer = LlamaTokenizer.from_pretrained(model_path, trust_remote_code=True)
     else:
         model = AutoModelForCausalLM.from_pretrained(model_path, load_in_low_bit=low_bit, trust_remote_code=True,
+                                                     optimize_model=optimize_model, max_output_len=max_output_len, max_prompt_len=int(in_out_len[0]), transpose_value_cache=True,
                                                      use_cache=True, attn_implementation="eager").eval()
         tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
     end = time.perf_counter()
@@ -643,6 +649,7 @@ def transformers_int4_npu_win(repo_id,
             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):
@@ -2016,12 +2023,15 @@ def run_pipeline_parallel_gpu(repo_id,
     streaming = False
     use_fp16_torch_dtype = False
     task = 'continuation'
+    optimize_model = False # only for transformers_int4_npu_win
     if 'streaming' in conf:
         streaming = conf['streaming']
     if 'use_fp16_torch_dtype' in conf:
         use_fp16_torch_dtype = conf['use_fp16_torch_dtype']
     if 'task' in conf:
         task = conf['task']
+    if 'optimize_model' in conf:
+        optimize_model = conf['optimize_model']
     lookahead = False
 
     import pandas as pd
@@ -2048,7 +2058,7 @@ def run_pipeline_parallel_gpu(repo_id,
                 if task in ['QA', 'summarize'] and conf['num_beams'] == 1 and batch_size == 1:
                     lookahead = True
                 run_model(model, api, in_out_pairs, conf['local_model_hub'], conf['warm_up'], conf['num_trials'], conf['num_beams'],
-                      conf['low_bit'], conf['cpu_embedding'], batch_size, streaming, use_fp16_torch_dtype, lookahead, task)
+                      conf['low_bit'], conf['cpu_embedding'], batch_size, streaming, use_fp16_torch_dtype, lookahead, task, optimize_model)
         df = pd.DataFrame(results, columns=['model', '1st token avg latency (ms)', '2+ avg latency (ms/token)', 'encoder time (ms)',
                                             'input/output tokens', 'batch_size', 'actual input/output tokens', 'num_beams', 'low_bit', 'cpu_embedding',
                                             'model loading time (s)', 'peak mem (GB)', 'streaming', 'use_fp16_torch_dtype'])

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

@@ -117,7 +117,7 @@ def from_pretrained(cls, *args, **kwargs):
         ignore_argument(kwargs, "pipeline_parallel_stages")
         optimize_model = kwargs.pop("optimize_model", False)
         max_output_len = kwargs.pop("max_output_len", 1024)
-        max_prompt_len = kwargs.pop("max_prompt_len", max_output_len)
+        max_prompt_len = kwargs.pop("max_prompt_len", 512)
         inter_pp = kwargs.pop("inter_pp", None)
         intra_pp = kwargs.pop("intra_pp", None)
         transpose_value_cache = kwargs.pop("transpose_value_cache", True)