Replace with IPEX-LLM in example comments (#10671)

* Replace with IPEX-LLM in example comments

* More replacement

* revert some changes

python/llm/example/CPU/Applications/streaming-llm/streaming_llm/utils.py

@@ -48,7 +48,7 @@ import ssl
 import urllib.request
 import os
 import json
-# code change to import from bigdl-llm API instead of using transformers API
+# code change to import from IPEX-LLM API instead of using transformers API
 from ipex_llm.transformers import AutoModelForCausalLM
 from transformers import LlamaTokenizer
 import intel_extension_for_pytorch as ipex

python/llm/example/CPU/Deepspeed-AutoTP/deepspeed_autotp.py

@@ -87,7 +87,7 @@ if __name__ == '__main__':
         replace_method="auto"
     )
 
-    # Apply BigDL-LLM INT4 optimizations on transformers
+    # Apply IPEX-LLM INT4 optimizations on transformers
     model = optimize_model(model.module.to(f'cpu'), low_bit='sym_int4')
 
     model = model.to(f'cpu:{local_rank}')
@@ -111,7 +111,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 do_sample=False,
                                 max_new_tokens=args.n_predict)

python/llm/example/CPU/HF-Transformers-AutoModels/Advanced-Quantizations/AWQ/generate.py

@@ -59,7 +59,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Advanced-Quantizations/GGUF/generate.py

@@ -44,7 +44,7 @@ if __name__ == '__main__':
 
     model_path = args.model
 
-    # Load gguf model and vocab, then convert them to bigdl-llm model and huggingface tokenizer
+    # Load gguf model and vocab, then convert them to IPEX-LLM model and huggingface tokenizer
     model, tokenizer = AutoModelForCausalLM.from_gguf(model_path, low_bit = args.low_bit,)
 
     # Generate predicted tokens

python/llm/example/CPU/HF-Transformers-AutoModels/Advanced-Quantizations/GPTQ/generate.py

@@ -60,7 +60,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/aquila/generate.py

@@ -56,7 +56,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/aquila2/generate.py

@@ -56,7 +56,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/baichuan/generate.py

@@ -56,7 +56,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/baichuan2/generate.py

@@ -45,7 +45,7 @@ if __name__ == '__main__':
     # if your selected model is capable of utilizing previous key/value attentions
     # to enhance decoding speed, but has `"use_cache": false` in its model config,
     # it is important to set `use_cache=True` explicitly in the `generate` function
-    # to obtain optimal performance with BigDL-LLM INT4 optimizations
+    # to obtain optimal performance with IPEX-LLM INT4 optimizations
     model = AutoModelForCausalLM.from_pretrained(model_path,
                                                  load_in_4bit=True,
                                                  trust_remote_code=True,

python/llm/example/CPU/HF-Transformers-AutoModels/Model/bluelm/generate.py

@@ -56,7 +56,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/chatglm/generate.py

@@ -57,7 +57,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/chatglm2/generate.py

@@ -57,7 +57,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/chatglm3/generate.py

@@ -57,7 +57,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/codellama/generate.py

@@ -55,7 +55,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/codeshell/generate.py

@@ -56,7 +56,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
 
         output = model.generate(input_ids, max_new_tokens=args.n_predict)
 

python/llm/example/CPU/HF-Transformers-AutoModels/Model/deepseek/generate.py

@@ -61,7 +61,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/dolly_v1/generate.py

@@ -61,7 +61,7 @@ if __name__ == '__main__':
         st = time.time()
         # enabling `use_cache=True` allows the model to utilize the previous
         # key/values attentions to speed up decoding;
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations,
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations,
         # it is important to set use_cache=True for Dolly v1 models
         output = model.generate(input_ids,
                                 use_cache=True,

python/llm/example/CPU/HF-Transformers-AutoModels/Model/dolly_v2/generate.py

@@ -64,7 +64,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict,
                                 pad_token_id=tokenizer.pad_token_id,

python/llm/example/CPU/HF-Transformers-AutoModels/Model/falcon/generate.py

@@ -57,7 +57,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/flan-t5/generate.py

@@ -60,7 +60,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/fuyu/generate.py

@@ -38,7 +38,7 @@ if __name__ == '__main__':
     image = Image.open(args.image_path)
 
     # Load model
-    # For successful BigDL-LLM optimization on Fuyu, skip the 'vision_embed_tokens' module during optimization
+    # For successful IPEX-LLM optimization on Fuyu, skip the 'vision_embed_tokens' module during optimization
     model = AutoModelForCausalLM.from_pretrained(model_path, device_map='cpu',
                                                  load_in_4bit = True,
                                                  trust_remote_code=True,

python/llm/example/CPU/HF-Transformers-AutoModels/Model/gemma/generate.py

@@ -61,7 +61,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/internlm-xcomposer/chat.py

@@ -37,7 +37,7 @@ if __name__ == '__main__':
     image = args.image_path
 
     # Load model
-    # For successful BigDL-LLM optimization on InternLM-XComposer, skip the 'qkv' module during optimization
+    # For successful IPEX-LLM optimization on InternLM-XComposer, skip the 'qkv' module during optimization
     model = AutoModelForCausalLM.from_pretrained(model_path, device='cpu', load_in_4bit=True,
                                                  trust_remote_code=True, modules_to_not_convert=['qkv'])
 

python/llm/example/CPU/HF-Transformers-AutoModels/Model/internlm/generate.py

@@ -57,7 +57,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/internlm2/generate.py

@@ -56,7 +56,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/llama2/generate.py

@@ -60,7 +60,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/mistral/generate.py

@@ -55,7 +55,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/mixtral/generate.py

@@ -61,7 +61,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/moss/generate.py

@@ -56,7 +56,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/mpt/generate.py

@@ -55,7 +55,7 @@ if __name__ == '__main__':
         input_ids = tokenizer.encode(prompt, return_tensors="pt")
         # enabling `use_cache=True` allows the model to utilize the previous
         # key/values attentions to speed up decoding;
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations,
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations,
         # it is important to set use_cache=True for MPT models
         mpt_generation_config = GenerationConfig(
             max_new_tokens=args.n_predict, 

python/llm/example/CPU/HF-Transformers-AutoModels/Model/phi-1_5/generate.py

@@ -58,7 +58,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
 
         # Note that phi-1_5 uses GenerationConfig to enable 'use_cache'
         output = model.generate(input_ids, do_sample=False, max_new_tokens=args.n_predict, generation_config = generation_config)

python/llm/example/CPU/HF-Transformers-AutoModels/Model/phi-2/generate.py

@@ -58,7 +58,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
 
         model.generation_config.pad_token_id = model.generation_config.eos_token_id
         # Note that phi-2 uses GenerationConfig to enable 'use_cache'

python/llm/example/CPU/HF-Transformers-AutoModels/Model/phixtral/generate.py

@@ -58,7 +58,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
 
         # Note that phixtral uses GenerationConfig to enable 'use_cache'
         output = model.generate(input_ids, do_sample=False, max_new_tokens=args.n_predict, generation_config = generation_config)

python/llm/example/CPU/HF-Transformers-AutoModels/Model/phoenix/generate.py

@@ -55,7 +55,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 do_sample=False,
                                 max_new_tokens=args.n_predict)

python/llm/example/CPU/HF-Transformers-AutoModels/Model/qwen-vl/chat.py

@@ -36,7 +36,7 @@ if __name__ == '__main__':
     model_path = args.repo_id_or_model_path  
         
     # Load model
-    # For successful BigDL-LLM optimization on Qwen-VL-Chat, skip the 'c_fc' and 'out_proj' modules during optimization
+    # For successful IPEX-LLM optimization on Qwen-VL-Chat, skip the 'c_fc' and 'out_proj' modules during optimization
     model = AutoModelForCausalLM.from_pretrained(model_path, 
                                                  load_in_4bit=True, 
                                                  device_map="cpu", 

python/llm/example/CPU/HF-Transformers-AutoModels/Model/qwen/generate.py

@@ -64,7 +64,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/redpajama/generate.py

@@ -56,7 +56,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(**inputs,
                                 max_new_tokens=args.n_predict,
                                 do_sample=True,

python/llm/example/CPU/HF-Transformers-AutoModels/Model/replit/generate.py

@@ -55,7 +55,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/skywork/generate.py

@@ -55,7 +55,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/solar/generate.py

@@ -59,7 +59,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/starcoder/generate.py

@@ -55,7 +55,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/vicuna/generate.py

@@ -53,7 +53,7 @@ if __name__ == '__main__':
         st = time.time()
         # enabling `use_cache=True` allows the model to utilize the previous
         # key/values attentions to speed up decoding;
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations,
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations,
         # it is important to set use_cache=True for vicuna-v1.3 models
         output = model.generate(input_ids,
                                 use_cache=True,

python/llm/example/CPU/HF-Transformers-AutoModels/Model/whisper/recognize.py

@@ -63,7 +63,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         predicted_ids = model.generate(input_features,
                                        forced_decoder_ids=forced_decoder_ids)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/wizardcoder-python/generate.py

@@ -58,7 +58,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/yi/generate.py

@@ -61,7 +61,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/HF-Transformers-AutoModels/Model/yuan2/generate.py

@@ -57,7 +57,7 @@ if __name__ == '__main__':
     # if your selected model is capable of utilizing previous key/value attentions
     # to enhance decoding speed, but has `"use_cache": false` in its model config,
     # it is important to set `use_cache=True` explicitly in the `generate` function
-    # to obtain optimal performance with BigDL-LLM INT4 optimizations
+    # to obtain optimal performance with IPEX-LLM INT4 optimizations
     outputs = model.generate(inputs, do_sample=True, top_k=5, max_length=args.n_predict)
     end_time = time.time()
 

python/llm/example/CPU/HF-Transformers-AutoModels/Model/ziya/generate.py

@@ -42,7 +42,7 @@ if __name__ == '__main__':
     from ipex_llm.transformers import AutoModelForCausalLM
     # enabling `use_cache=True` allows the model to utilize the previous
     # key/values attentions to speed up decoding;
-    # to obtain optimal performance with BigDL-LLM INT4 optimizations,
+    # to obtain optimal performance with IPEX-LLM INT4 optimizations,
     # it is important to set use_cache=True for Ziya models
     model = AutoModelForCausalLM.from_pretrained(model_path,
                                                  load_in_4bit=True,

python/llm/example/CPU/LlamaIndex/rag.py

@@ -163,7 +163,7 @@ def messages_to_prompt(messages):
 def main(args):
     embed_model = HuggingFaceEmbedding(model_name=args.embedding_model_path)
     
-    # Use custom LLM in BigDL
+    # Use custom LLM in IPEX-LLM
     from ipex_llm.llamaindex.llms import BigdlLLM
     llm = BigdlLLM(
         model_name=args.model_path,

python/llm/example/CPU/ModelScope-Models/generate.py

@@ -59,7 +59,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/Native-Models/native_int4_pipeline.py

@@ -56,8 +56,8 @@ def load(model_path, model_family, n_threads):
 def inference(llm, repo_id_or_model_path, model_family, prompt):
 
     if model_family in ['llama', 'gptneox', 'bloom', 'starcoder', 'chatglm']:
-        # ------ Option 1: Use bigdl-llm based tokenizer
-        print('-'*20, ' bigdl-llm based tokenizer ', '-'*20)
+        # ------ Option 1: Use IPEX-LLM based tokenizer
+        print('-'*20, ' IPEX-LLM based tokenizer ', '-'*20)
         st = time.time()
 
         # please note that the prompt here can either be a string or a list of string
@@ -126,13 +126,13 @@ def main():
     if args.model_family == 'llama2':
         args.model_family = 'llama'
 
-    # Step 1: convert original model to BigDL llm model
-    bigdl_llm_path = convert(repo_id_or_model_path=repo_id_or_model_path,
+    # Step 1: convert original model to IPEX-LLM model
+    ipex_llm_path = convert(repo_id_or_model_path=repo_id_or_model_path,
                              model_family=args.model_family,
                              tmp_path=args.tmp_path)
     
     # Step 2: load int4 model
-    llm = load(model_path=bigdl_llm_path,
+    llm = load(model_path=ipex_llm_path,
                model_family=args.model_family,
                n_threads=args.thread_num)
 

python/llm/example/CPU/PyTorch-Models/Model/aquila2/generate.py

@@ -44,7 +44,7 @@ if __name__ == '__main__':
                                                  torch_dtype='auto',
                                                  low_cpu_mem_usage=True)
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/bark/synthesize_speech.py

@@ -38,7 +38,7 @@ if __name__ == '__main__':
     model = Bark.init_from_config(config)
     model.load_checkpoint(config, checkpoint_dir=model_path, eval=True)
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Synthesize speech with the given input

python/llm/example/CPU/PyTorch-Models/Model/bert/extract_feature.py

@@ -38,7 +38,7 @@ if __name__ == '__main__':
                                       torch_dtype="auto",
                                       low_cpu_mem_usage=True)
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/bluelm/generate.py

@@ -40,7 +40,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
     
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/chatglm/generate.py

@@ -41,7 +41,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModel.from_pretrained(model_path, trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
     
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/chatglm3/generate.py

@@ -41,7 +41,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModel.from_pretrained(model_path, trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
     
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/codellama/generate.py

@@ -41,7 +41,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True)
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/codeshell/generate.py

@@ -40,7 +40,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
     
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/deciLM-7b/generate.py

@@ -50,7 +50,7 @@ if __name__ == '__main__':
         trust_remote_code=True,
     )
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/deepseek/generate.py

@@ -46,7 +46,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True)
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/flan-t5/generate.py

@@ -40,7 +40,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForSeq2SeqLM.from_pretrained(model_path, trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     # "wo" module is not converted due to some issues of T5 model
     # (https://github.com/huggingface/transformers/issues/20287),
     # "lm_head" module is not converted to generate outputs with better quality

python/llm/example/CPU/PyTorch-Models/Model/fuyu/generate.py

@@ -40,8 +40,8 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, device_map='cpu', trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
-    # For successful BigDL-LLM optimization on Fuyu, skip the 'vision_embed_tokens' module during optimization
+    # With only one line to enable IPEX-LLM optimization on model
+    # For successful IPEX-LLM optimization on Fuyu, skip the 'vision_embed_tokens' module during optimization
     model = optimize_model(model,
                            low_bit='sym_int4',
                            modules_to_not_convert=['vision_embed_tokens'])

python/llm/example/CPU/PyTorch-Models/Model/internlm-xcomposer/chat.py

@@ -38,8 +38,8 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, device='cpu', trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
-    # For successful BigDL-LLM optimization on InternLM-XComposer, skip the 'qkv' module during optimization
+    # With only one line to enable IPEX-LLM optimization on model
+    # For successful IPEX-LLM optimization on InternLM-XComposer, skip the 'qkv' module during optimization
     model = optimize_model(model,
                            low_bit='sym_int4',
                            modules_to_not_convert=['qkv'])

python/llm/example/CPU/PyTorch-Models/Model/llama2/generate.py

@@ -45,7 +45,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True)
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/llava/generate.py

@@ -292,7 +292,7 @@ if __name__ == '__main__':
                                                                  model_base=None,
                                                                  model_name=model_name)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Generate image tensor

python/llm/example/CPU/PyTorch-Models/Model/mamba/generate.py

@@ -42,7 +42,7 @@ if __name__ == '__main__':
     # Load model
     model = MambaLMHeadModel.from_pretrained(model_path)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model, low_bit='asym_int4', modules_to_not_convert=["dt_proj", "x_proj", "out_proj"])
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/mistral/generate.py

@@ -44,7 +44,7 @@ if __name__ == '__main__':
                                                  torch_dtype='auto',
                                                  low_cpu_mem_usage=True)
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/mixtral/generate.py

@@ -44,7 +44,7 @@ if __name__ == '__main__':
                                                  torch_dtype='auto',
                                                  low_cpu_mem_usage=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/openai-whisper/recognize.py

@@ -46,7 +46,7 @@ if __name__ == '__main__':
     # Load whisper model under pytorch framework
     model = whisper.load_model(args.model_name)
 
-    # With only one line to enable bigdl optimize on a pytorch model
+    # With only one line to enable IPEX-LLM optimize on a pytorch model
     model = optimize_model(model)
 
     st = time.time()

python/llm/example/CPU/PyTorch-Models/Model/phi-1_5/generate.py

@@ -42,7 +42,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
     
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/phi-2/generate.py

@@ -42,7 +42,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
     
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/qwen-vl/chat.py

@@ -37,8 +37,8 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, device_map="cpu", trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
-    # For successful BigDL-LLM optimization on Qwen-VL-Chat, skip the 'c_fc' and 'out_proj' modules during optimization
+    # With only one line to enable IPEX-LLM optimization on model
+    # For successful IPEX-LLM optimization on Qwen-VL-Chat, skip the 'c_fc' and 'out_proj' modules during optimization
     model = optimize_model(model, 
                            low_bit='sym_int4', 
                            modules_to_not_convert=['c_fc', 'out_proj'])

python/llm/example/CPU/PyTorch-Models/Model/skywork/generate.py

@@ -41,7 +41,7 @@ if __name__ == '__main__':
     model = AutoModelForCausalLM.from_pretrained(model_path,
                                                  trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/solar/generate.py

@@ -45,7 +45,7 @@ if __name__ == '__main__':
                                                  torch_dtype=torch.float16,
                                                  trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
     
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/wizardcoder-python/generate.py

@@ -44,7 +44,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True)
 
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     # Load tokenizer

python/llm/example/CPU/PyTorch-Models/Model/yi/generate.py

@@ -54,7 +54,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/CPU/PyTorch-Models/Model/yuan2/generate.py

@@ -48,7 +48,7 @@ if __name__ == '__main__':
     print("Creating model...")
     model = AutoModelForCausalLM.from_pretrained(model_path, device_map="cpu", trust_remote_code=True, torch_dtype=torch.float16).eval()
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = optimize_model(model)
 
     prompt = YUAN2_PROMPT_FORMAT.format(prompt=args.prompt)
@@ -59,7 +59,7 @@ if __name__ == '__main__':
     # if your selected model is capable of utilizing previous key/value attentions
     # to enhance decoding speed, but has `"use_cache": false` in its model config,
     # it is important to set `use_cache=True` explicitly in the `generate` function
-    # to obtain optimal performance with BigDL-LLM INT4 optimizations
+    # to obtain optimal performance with IPEX-LLM INT4 optimizations
     outputs = model.generate(inputs, do_sample=True, top_k=5, max_length=args.n_predict)
     end_time = time.time()
 

python/llm/example/CPU/PyTorch-Models/Model/ziya/generate.py

@@ -42,7 +42,7 @@ if __name__ == '__main__':
     from ipex_llm import optimize_model
     # enabling `use_cache=True` allows the model to utilize the previous
     # key/values attentions to speed up decoding;
-    # to obtain optimal performance with BigDL-LLM `optimization_model` API optimizations,
+    # to obtain optimal performance with IPEX-LLM `optimization_model` API optimizations,
     # it is important to set use_cache=True for Ziya models
     model = AutoModelForCausalLM.from_pretrained(model_path,
                                                  trust_remote_code=True,

python/llm/example/CPU/PyTorch-Models/More-Data-Types/generate.py

@@ -49,7 +49,7 @@ if __name__ == '__main__':
     # Load model
     model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True)
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     # `low_bit` param support `sym_int4`, `asym_int4`, `sym_int5`, `asym_int5` and `sym_int8`
     # By specifying `low_bit` param, relevant low bit optimizations will be applied to the model
     model = optimize_model(model, low_bit=low_bit)

python/llm/example/CPU/QLoRA-FineTuning/alpaca-qlora/alpaca_qlora_finetuning_cpu.py

@@ -49,7 +49,7 @@ from utils.prompter import Prompter
 from transformers import BitsAndBytesConfig
 from ipex_llm.transformers import AutoModelForCausalLM
 
-# import them from ipex_llm.transformers.qlora to get a BigDL-LLM compatible Peft model
+# import them from ipex_llm.transformers.qlora to get a IPEX-LLM compatible Peft model
 from ipex_llm.transformers.qlora import get_peft_model, prepare_model_for_kbit_training, LoraConfig
 from ipex_llm.utils.isa_checker import ISAChecker
 
@@ -64,7 +64,7 @@ def get_int_from_env(env_keys, default):
 def train(
     # model/data params
     base_model: str = "meta-llama/Llama-2-7b-hf",  # the only required argument, default to be "meta-llama/Llama-2-7b-hf"
-    saved_low_bit_model: str = None,  # optional, the path to the saved model with bigdl-llm low-bit optimization
+    saved_low_bit_model: str = None,  # optional, the path to the saved model with ipex-llm low-bit optimization
     data_path: str = "yahma/alpaca-cleaned",
     output_dir: str = "./bigdl-qlora-alpaca",
     # training hyperparams
@@ -256,7 +256,7 @@ def train(
             ]  # could be sped up, probably
         return tokenized_full_prompt
 
-    # Prepare a BigDL-LLM compatible Peft model
+    # Prepare a IPEX-LLM compatible Peft model
     model = prepare_model_for_kbit_training(model, use_gradient_checkpointing=gradient_checkpointing)
 
     config = LoraConfig(

python/llm/example/GPU/Deepspeed-AutoTP/deepspeed_autotp.py

@@ -63,7 +63,7 @@ if __name__ == '__main__':
     low_bit = args.low_bit
 
     # First use CPU as accelerator
-    # Convert to deepspeed model and apply bigdl-llm optimization on CPU to decrease GPU memory usage
+    # Convert to deepspeed model and apply IPEX-LLM optimization on CPU to decrease GPU memory usage
     current_accel = CPU_Accelerator()
     set_accelerator(current_accel)
     model = AutoModelForCausalLM.from_pretrained(args.repo_id_or_model_path,
@@ -80,7 +80,7 @@ if __name__ == '__main__':
         replace_method="auto",
     )
 
-    # Use bigdl-llm `optimize_model` to convert the model into optimized low bit format
+    # Use IPEX-LLM `optimize_model` to convert the model into optimized low bit format
     # Convert the rest of the model into float16 to reduce allreduce traffic
     model = optimize_model(model.module.to(f'cpu'), low_bit=low_bit).to(torch.float16)
 
@@ -119,7 +119,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 do_sample=False,
                                 max_new_tokens=args.n_predict)

python/llm/example/GPU/HF-Transformers-AutoModels/Advanced-Quantizations/AWQ/generate.py

@@ -58,7 +58,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/GPU/HF-Transformers-AutoModels/Advanced-Quantizations/GGUF-IQ2/generate.py

@@ -71,7 +71,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM Low Bit optimizations
+        # to obtain optimal performance with IPEX-LLM Low Bit optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict,
                                 repetition_penalty=1.1)

python/llm/example/GPU/HF-Transformers-AutoModels/Advanced-Quantizations/GGUF/generate.py

@@ -44,7 +44,7 @@ if __name__ == '__main__':
 
     model_path = args.model
 
-    # Load gguf model and vocab, then convert them to bigdl-llm model and huggingface tokenizer
+    # Load gguf model and vocab, then convert them to IPEX-LLM model and huggingface tokenizer
     model, tokenizer = AutoModelForCausalLM.from_gguf(model_path)
     model = model.to('xpu')
 

python/llm/example/GPU/HF-Transformers-AutoModels/Advanced-Quantizations/GPTQ/generate.py

@@ -59,7 +59,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         end = time.time()

python/llm/example/GPU/HF-Transformers-AutoModels/Model/aquila/generate.py

@@ -60,7 +60,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         torch.xpu.synchronize()

python/llm/example/GPU/HF-Transformers-AutoModels/Model/aquila2/generate.py

@@ -60,7 +60,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         torch.xpu.synchronize()

python/llm/example/GPU/HF-Transformers-AutoModels/Model/baichuan/generate.py

@@ -64,7 +64,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         torch.xpu.synchronize()

python/llm/example/GPU/HF-Transformers-AutoModels/Model/baichuan2/generate.py

@@ -44,7 +44,7 @@ if __name__ == '__main__':
     # if your selected model is capable of utilizing previous key/value attentions
     # to enhance decoding speed, but has `"use_cache": false` in its model config,
     # it is important to set `use_cache=True` explicitly in the `generate` function
-    # to obtain optimal performance with BigDL-LLM INT4 optimizations
+    # to obtain optimal performance with IPEX-LLM INT4 optimizations
     # When running LLMs on Intel iGPUs for Windows users, we recommend setting `cpu_embedding=True` in the from_pretrained function.
     # This will allow the memory-intensive embedding layer to utilize the CPU instead of iGPU.
     model = AutoModelForCausalLM.from_pretrained(model_path,

python/llm/example/GPU/HF-Transformers-AutoModels/Model/bluelm/generate.py

@@ -64,7 +64,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         torch.xpu.synchronize()

python/llm/example/GPU/HF-Transformers-AutoModels/Model/chatglm2/generate.py

@@ -67,7 +67,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         torch.xpu.synchronize()

python/llm/example/GPU/HF-Transformers-AutoModels/Model/chatglm3/generate.py

@@ -67,7 +67,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         torch.xpu.synchronize()

python/llm/example/GPU/HF-Transformers-AutoModels/Model/chinese-llama2/generate.py

@@ -57,7 +57,7 @@ if __name__ == '__main__':
     # if your selected model is capable of utilizing previous key/value attentions
     # to enhance decoding speed, but has `"use_cache": false` in its model config,
     # it is important to set `use_cache=True` explicitly to obtain optimal
-    # performance with BigDL-LLM INT4 optimizations
+    # performance with IPEX-LLM INT4 optimizations
     # When running LLMs on Intel iGPUs for Windows users, we recommend setting `cpu_embedding=True` in the from_pretrained function.
     # This will allow the memory-intensive embedding layer to utilize the CPU instead of iGPU.
     model = AutoModelForCausalLM.from_pretrained(model_path,

python/llm/example/GPU/HF-Transformers-AutoModels/Model/codellama/generate.py

@@ -67,7 +67,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         torch.xpu.synchronize()

python/llm/example/GPU/HF-Transformers-AutoModels/Model/deciLM-7b/generate.py

@@ -55,7 +55,7 @@ if __name__ == '__main__':
         cpu_embedding=True
     )
     
-    # With only one line to enable BigDL-LLM optimization on model
+    # With only one line to enable IPEX-LLM optimization on model
     model = model.to('xpu')
     # Load tokenizer
     tokenizer = AutoTokenizer.from_pretrained(model_path)

python/llm/example/GPU/HF-Transformers-AutoModels/Model/dolly-v1/generate.py

@@ -64,7 +64,7 @@ if __name__ == '__main__':
         st = time.time()
         # enabling `use_cache=True` allows the model to utilize the previous
         # key/values attentions to speed up decoding;
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations,
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations,
         # it is important to set use_cache=True for Dolly v1 models
         output = model.generate(input_ids,
                                 use_cache=True,

python/llm/example/GPU/HF-Transformers-AutoModels/Model/dolly-v2/generate.py

@@ -67,7 +67,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict,
                                 pad_token_id=tokenizer.pad_token_id,

python/llm/example/GPU/HF-Transformers-AutoModels/Model/falcon/generate.py

@@ -67,7 +67,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         torch.xpu.synchronize()

python/llm/example/GPU/HF-Transformers-AutoModels/Model/flan-t5/generate.py

@@ -69,7 +69,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         torch.xpu.synchronize()

python/llm/example/GPU/HF-Transformers-AutoModels/Model/gemma/generate.py

@@ -68,7 +68,7 @@ if __name__ == '__main__':
         # if your selected model is capable of utilizing previous key/value attentions
         # to enhance decoding speed, but has `"use_cache": false` in its model config,
         # it is important to set `use_cache=True` explicitly in the `generate` function
-        # to obtain optimal performance with BigDL-LLM INT4 optimizations
+        # to obtain optimal performance with IPEX-LLM INT4 optimizations
         output = model.generate(input_ids,
                                 max_new_tokens=args.n_predict)
         torch.xpu.synchronize()