[NPU L0] Update streaming mode of example (#12312)

python/llm/example/NPU/HF-Transformers-AutoModels/LLM/Pipeline-Models/README.md

@@ -72,28 +72,21 @@ Arguments info:
 - `--max-context-len MAX_CONTEXT_LEN`: Defines the maximum sequence length for both input and output tokens. It is default to be `1024`.
 - `--max-prompt-len MAX_PROMPT_LEN`: Defines the maximum number of tokens that the input prompt can contain. It is default to be `512`.
 - `--disable-transpose-value-cache`: Disable the optimization of transposing value cache.
+- `--disable-streaming`: Disable streaming mode of generation.
 
-### Sample Output
+### Sample Output of Streaming Mode
 #### [meta-llama/Llama-2-7b-chat-hf](https://huggingface.co/meta-llama/Llama-2-7b-chat-hf)
  
 ```log
- Number of input tokens: 28
- Generated tokens: 32
- First token generation time: xxxx s
- Generation average latency: xxxx ms, (xxxx token/s)
- Generation time: xxxx s
-
-Inference time: xxxx s
 -------------------- Input --------------------
-<s><s> [INST] <<SYS>>
+input length: 28
+<s>[INST] <<SYS>>
 
 <</SYS>>
 
 What is AI? [/INST]
 -------------------- Output --------------------
-<s><s> [INST] <<SYS>>
+ AI (Artificial Intelligence) is a field of computer science and technology that focuses on the development of intelligent machines that can perform
 
-<</SYS>>
+Inference time: xxxx s
-
-What is AI? [/INST]  AI (Artificial Intelligence) is a field of computer science and technology that focuses on the development of intelligent machines that can perform
 ```

python/llm/example/NPU/HF-Transformers-AutoModels/LLM/Pipeline-Models/baichuan2.py

@@ -20,7 +20,7 @@ import torch
 import time
 import argparse
 from ipex_llm.transformers.npu_model import AutoModelForCausalLM
-from transformers import AutoTokenizer
+from transformers import AutoTokenizer, TextStreamer
 from transformers.utils import logging
 
 logger = logging.get_logger(__name__)
@@ -61,6 +61,7 @@ if __name__ == "__main__":
     parser.add_argument("--max-context-len", type=int, default=1024)
     parser.add_argument("--max-prompt-len", type=int, default=512)
     parser.add_argument("--disable-transpose-value-cache", action="store_true", default=False)
+    parser.add_argument("--disable-streaming", action="store_true", default=False)
 
     args = parser.parse_args()
     model_path = args.repo_id_or_model_path
@@ -92,6 +93,11 @@ if __name__ == "__main__":
     if args.lowbit_path and not os.path.exists(args.lowbit_path):
         model.save_low_bit(args.lowbit_path)
 
+    if args.disable_streaming:
+        streamer = None
+    else:
+        streamer = TextStreamer(tokenizer=tokenizer, skip_special_tokens=True)
+
     DEFAULT_SYSTEM_PROMPT = """\
     """
 
@@ -99,22 +105,22 @@ if __name__ == "__main__":
     print("done")
     with torch.inference_mode():
         print("finish to load")
-        for i in range(5):
+        for i in range(3):
             prompt = get_prompt(args.prompt, [], system_prompt=DEFAULT_SYSTEM_PROMPT)
             _input_ids = tokenizer.encode(prompt, return_tensors="pt")
+            print("-" * 20, "Input", "-" * 20)
             print("input length:", len(_input_ids[0]))
+            print(prompt)
+            print("-" * 20, "Output", "-" * 20)
             st = time.time()
             output = model.generate(
-                _input_ids, max_new_tokens=args.n_predict, do_print=True
+                _input_ids, max_new_tokens=args.n_predict, streamer=streamer
             )
             end = time.time()
-            print(f"Inference time: {end-st} s")
+            if args.disable_streaming:
-            input_str = tokenizer.decode(_input_ids[0], skip_special_tokens=False)
-            print("-" * 20, "Input", "-" * 20)
-            print(input_str)
                 output_str = tokenizer.decode(output[0], skip_special_tokens=False)
-            print("-" * 20, "Output", "-" * 20)
                 print(output_str)
+            print(f"Inference time: {end-st} s")
 
     print("-" * 80)
     print("done")

python/llm/example/NPU/HF-Transformers-AutoModels/LLM/Pipeline-Models/llama2.py

@@ -20,7 +20,7 @@ import torch
 import time
 import argparse
 from ipex_llm.transformers.npu_model import AutoModelForCausalLM
-from transformers import AutoTokenizer
+from transformers import AutoTokenizer, TextStreamer
 from transformers.utils import logging
 
 logger = logging.get_logger(__name__)
@@ -62,6 +62,7 @@ if __name__ == "__main__":
     parser.add_argument("--max-prompt-len", type=int, default=512)
     parser.add_argument("--quantization_group_size", type=int, default=0)
     parser.add_argument("--disable-transpose-value-cache", action="store_true", default=False)
+    parser.add_argument("--disable-streaming", action="store_true", default=False)
 
     args = parser.parse_args()
     model_path = args.repo_id_or_model_path
@@ -92,6 +93,11 @@ if __name__ == "__main__":
     if args.lowbit_path and not os.path.exists(args.lowbit_path):
         model.save_low_bit(args.lowbit_path)
     
+    if args.disable_streaming:
+        streamer = None
+    else:
+        streamer = TextStreamer(tokenizer=tokenizer, skip_special_tokens=True)
+
     DEFAULT_SYSTEM_PROMPT = """\
     """
 
@@ -99,22 +105,22 @@ if __name__ == "__main__":
     print("done")
     with torch.inference_mode():
         print("finish to load")
-        for i in range(5):
+        for i in range(3):
             prompt = get_prompt(args.prompt, [], system_prompt=DEFAULT_SYSTEM_PROMPT)
             _input_ids = tokenizer.encode(prompt, return_tensors="pt")
+            print("-" * 20, "Input", "-" * 20)
             print("input length:", len(_input_ids[0]))
+            print(prompt)
+            print("-" * 20, "Output", "-" * 20)
             st = time.time()
             output = model.generate(
-                _input_ids, max_new_tokens=args.n_predict, do_print=True
+                _input_ids, max_new_tokens=args.n_predict, streamer=streamer
             )
             end = time.time()
-            print(f"Inference time: {end-st} s")
+            if args.disable_streaming:
-            input_str = tokenizer.decode(_input_ids[0], skip_special_tokens=False)
-            print("-" * 20, "Input", "-" * 20)
-            print(input_str)
                 output_str = tokenizer.decode(output[0], skip_special_tokens=False)
-            print("-" * 20, "Output", "-" * 20)
                 print(output_str)
+            print(f"Inference time: {end-st} s")
 
     print("-" * 80)
     print("done")

python/llm/example/NPU/HF-Transformers-AutoModels/LLM/Pipeline-Models/llama3.py

@@ -20,7 +20,7 @@ import torch
 import time
 import argparse
 from ipex_llm.transformers.npu_model import AutoModelForCausalLM
-from transformers import AutoTokenizer
+from transformers import AutoTokenizer, TextStreamer
 from transformers.utils import logging
 
 logger = logging.get_logger(__name__)
@@ -68,6 +68,7 @@ if __name__ == "__main__":
     parser.add_argument("--max-prompt-len", type=int, default=512)
     parser.add_argument("--quantization_group_size", type=int, default=0)
     parser.add_argument("--disable-transpose-value-cache", action="store_true", default=False)
+    parser.add_argument("--disable-streaming", action="store_true", default=False)
 
     args = parser.parse_args()
     model_path = args.repo_id_or_model_path
@@ -98,26 +99,31 @@ if __name__ == "__main__":
     if args.lowbit_path and not os.path.exists(args.lowbit_path):
         model.save_low_bit(args.lowbit_path)
 
+    if args.disable_streaming:
+        streamer = None
+    else:
+        streamer = TextStreamer(tokenizer=tokenizer, skip_special_tokens=True)
+
     print("-" * 80)
     print("done")
     with torch.inference_mode():
         print("finish to load")
-        for i in range(5):
+        for i in range(3):
             prompt = get_prompt(args.prompt, [], system_prompt=DEFAULT_SYSTEM_PROMPT)
             _input_ids = tokenizer.encode(prompt, return_tensors="pt")
+            print("-" * 20, "Input", "-" * 20)
             print("input length:", len(_input_ids[0]))
+            print(prompt)
+            print("-" * 20, "Output", "-" * 20)
             st = time.time()
             output = model.generate(
-                _input_ids, max_new_tokens=args.n_predict, do_print=True
+                _input_ids, max_new_tokens=args.n_predict, streamer=streamer
             )
             end = time.time()
-            print(f"Inference time: {end-st} s")
+            if args.disable_streaming:
-            input_str = tokenizer.decode(_input_ids[0], skip_special_tokens=False)
-            print("-" * 20, "Input", "-" * 20)
-            print(input_str)
                 output_str = tokenizer.decode(output[0], skip_special_tokens=False)
-            print("-" * 20, "Output", "-" * 20)
                 print(output_str)
+            print(f"Inference time: {end-st} s")
 
     print("-" * 80)
     print("done")

python/llm/example/NPU/HF-Transformers-AutoModels/LLM/Pipeline-Models/minicpm.py

@@ -19,7 +19,7 @@ import torch
 import time
 import argparse
 from ipex_llm.transformers.npu_model import AutoModelForCausalLM
-from transformers import AutoTokenizer
+from transformers import AutoTokenizer, TextStreamer
 from transformers.utils import logging
 import os
 
@@ -48,6 +48,7 @@ if __name__ == "__main__":
     parser.add_argument("--max-context-len", type=int, default=1024)
     parser.add_argument("--max-prompt-len", type=int, default=512)
     parser.add_argument("--disable-transpose-value-cache", action="store_true", default=False)
+    parser.add_argument("--disable-streaming", action="store_true", default=False)
 
     args = parser.parse_args()
     model_path = args.repo_id_or_model_path
@@ -79,26 +80,31 @@ if __name__ == "__main__":
     if args.lowbit_path and not os.path.exists(args.lowbit_path):
         model.save_low_bit(args.lowbit_path)
 
+    if args.disable_streaming:
+        streamer = None
+    else:
+        streamer = TextStreamer(tokenizer=tokenizer, skip_special_tokens=True)
+
     print("-" * 80)
     print("done")
     with torch.inference_mode():
         print("finish to load")
-        for i in range(5):
+        for i in range(3):
             prompt = "<用户>{}<AI>".format(args.prompt)
             _input_ids = tokenizer.encode(prompt, return_tensors="pt")
+            print("-" * 20, "Input", "-" * 20)
             print("input length:", len(_input_ids[0]))
+            print(prompt)
+            print("-" * 20, "Output", "-" * 20)
             st = time.time()
             output = model.generate(
-                _input_ids, max_new_tokens=args.n_predict, do_print=True
+                _input_ids, max_new_tokens=args.n_predict, streamer=streamer
             )
             end = time.time()
-            print(f"Inference time: {end-st} s")
+            if args.disable_streaming:
-            input_str = tokenizer.decode(_input_ids[0], skip_special_tokens=False)
-            print("-" * 20, "Input", "-" * 20)
-            print(input_str)
                 output_str = tokenizer.decode(output[0], skip_special_tokens=False)
-            print("-" * 20, "Output", "-" * 20)
                 print(output_str)
+            print(f"Inference time: {end-st} s")
 
     print("-" * 80)
     print("done")

python/llm/example/NPU/HF-Transformers-AutoModels/LLM/Pipeline-Models/qwen.py

@@ -20,7 +20,7 @@ import torch
 import time
 import argparse
 from ipex_llm.transformers.npu_model import AutoModelForCausalLM
-from transformers import AutoTokenizer
+from transformers import AutoTokenizer, TextStreamer
 from transformers.utils import logging
 
 logger = logging.get_logger(__name__)
@@ -50,6 +50,7 @@ if __name__ == "__main__":
     parser.add_argument('--load_in_low_bit', type=str, default="sym_int4",
                         help='Load in low bit to use')
     parser.add_argument("--disable-transpose-value-cache", action="store_true", default=False)
+    parser.add_argument("--disable-streaming", action="store_true", default=False)
 
     args = parser.parse_args()
     model_path = args.repo_id_or_model_path
@@ -81,6 +82,11 @@ if __name__ == "__main__":
     if args.lowbit_path and not os.path.exists(args.lowbit_path):
         model.save_low_bit(args.lowbit_path)
 
+    if args.disable_streaming:
+        streamer = None
+    else:
+        streamer = TextStreamer(tokenizer=tokenizer, skip_special_tokens=True)
+
     print("-" * 80)
     print("done")
     messages = [{"role": "system", "content": "You are a helpful assistant."},
@@ -90,21 +96,21 @@ if __name__ == "__main__":
                                          add_generation_prompt=True)
     with torch.inference_mode():
         print("finish to load")
-        for i in range(5):
+        for i in range(3):
             _input_ids = tokenizer([text], return_tensors="pt").input_ids
+            print("-" * 20, "Input", "-" * 20)
             print("input length:", len(_input_ids[0]))
+            print(text)
+            print("-" * 20, "Output", "-" * 20)
             st = time.time()
             output = model.generate(
-                _input_ids, max_new_tokens=args.n_predict, do_print=True
+                _input_ids, max_new_tokens=args.n_predict, streamer=streamer
             )
             end = time.time()
-            print(f"Inference time: {end-st} s")
+            if args.disable_streaming:
-            input_str = tokenizer.decode(_input_ids[0], skip_special_tokens=False)
-            print("-" * 20, "Input", "-" * 20)
-            print(input_str)
                 output_str = tokenizer.decode(output[0], skip_special_tokens=False)
-            print("-" * 20, "Output", "-" * 20)
                 print(output_str)
+            print(f"Inference time: {end-st} s")
 
     print("-" * 80)
     print("done")

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

@@ -134,7 +134,6 @@ def generate(
             try:
                 input_pipe = open(in_pipe_path, "wb")
             except:
-                print('Waiting for input pipe')
                 time.sleep(1)
             else:
                 break
@@ -143,7 +142,6 @@ def generate(
             try:
                 output_pipe = open(out_pipe_path, "rb")
             except:
-                print('Waiting for output pipe')
                 time.sleep(1)
             else:
                 break
@@ -152,7 +150,7 @@ def generate(
 
         bdata = str.encode(str(temp_dir))
         invalidInputError(len(bdata) <= 2000,
-                          f"Leng of input directory is too long ({len(bdata)}), "
+                          f"Length of input directory is too long ({len(bdata)}), "
                           "which may cause read error.")
         input_pipe.write(bdata)
         input_pipe.flush()