Update README.md

README.md

@@ -45,7 +45,7 @@ sampling_params = SamplingParams(temperature=0.8, top_p=0.95)
 
 if __name__ == '__main__':
     # Create an LLM.
-    llm = LLM(model="pytorch/Phi-4-mini-instruct-int4wo-hqq")
+    llm = LLM(model="pytorch/Phi-4-mini-instruct-INT4")
     # Generate texts from the prompts.
     # The output is a list of RequestOutput objects
     # that contain the prompt, generated text, and other information.
@@ -68,7 +68,7 @@ this is expected be resolved in pytorch 2.8.
 ## Serving
 Then we can serve with the following command:
 ```Shell
-vllm serve pytorch/Phi-4-mini-instruct-int4wo-hqq --tokenizer microsoft/Phi-4-mini-instruct -O3
+vllm serve pytorch/Phi-4-mini-instruct-INT4 --tokenizer microsoft/Phi-4-mini-instruct -O3
 ```
 
 
@@ -89,7 +89,7 @@ from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
  
 torch.random.manual_seed(0)
 
-model_path = "pytorch/Phi-4-mini-instruct-int4wo-hqq"
+model_path = "pytorch/Phi-4-mini-instruct-INT4"
 
 model = AutoModelForCausalLM.from_pretrained(
     model_path,
@@ -149,7 +149,7 @@ tokenizer = AutoTokenizer.from_pretrained(model_id)
 # Push to hub
 USER_ID = "YOUR_USER_ID"
 MODEL_NAME = model_id.split("/")[-1]
-save_to = f"{USER_ID}/{MODEL_NAME}-int4wo-hqq"
+save_to = f"{USER_ID}/{MODEL_NAME}-INT4"
 quantized_model.push_to_hub(save_to, safe_serialization=False)
 tokenizer.push_to_hub(save_to)
 
@@ -193,7 +193,7 @@ We rely on [lm-evaluation-harness](https://github.com/EleutherAI/lm-evaluation-h
 
 | Benchmark                        |                |                           |
 |----------------------------------|----------------|---------------------------|
-|                                  | Phi-4-mini-ins | Phi-4-mini-ins-int4wo-hqq | 
+|                                  | Phi-4-mini-ins | Phi-4-mini-ins-INT4 | 
 | **Popular aggregated benchmark** |                |                           |
 | mmlu (0-shot)                    | 66.73          |  63.56                    |
 | mmlu_pro (5-shot)                | 46.43          |  36.74                    |
@@ -224,9 +224,9 @@ https://github.com/EleutherAI/lm-evaluation-harness#install
 lm_eval --model hf --model_args pretrained=microsoft/Phi-4-mini-instruct --tasks hellaswag --device cuda:0 --batch_size 8
 ```
 
-## int4 weight only quantization with hqq (int4wo-hqq)
+## int4 weight only quantization with hqq (INT4)
 ```Shell
-lm_eval --model hf --model_args pretrained=pytorch/Phi-4-mini-instruct-int4wo-hqq --tasks hellaswag --device cuda:0 --batch_size 8
+lm_eval --model hf --model_args pretrained=pytorch/Phi-4-mini-instruct-INT4 --tasks hellaswag --device cuda:0 --batch_size 8
 ```
 </details>
 
@@ -236,7 +236,7 @@ lm_eval --model hf --model_args pretrained=pytorch/Phi-4-mini-instruct-int4wo-hq
 
 | Benchmark        |                |                                |
 |------------------|----------------|--------------------------------|
-|                  | Phi-4 mini-Ins | Phi-4-mini-instruct-int4wo-hqq | 
+|                  | Phi-4 mini-Ins | Phi-4-mini-instruct-INT4 | 
 | Peak Memory (GB) | 8.91           | 2.98 (67% reduction)           |
 
 
@@ -248,8 +248,8 @@ We can use the following code to get a sense of peak memory usage during inferen
 import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer, TorchAoConfig
 
-# use "microsoft/Phi-4-mini-instruct" or "pytorch/Phi-4-mini-instruct-int4wo-hqq"
-model_id = "pytorch/Phi-4-mini-instruct-int4wo-hqq"
+# use "microsoft/Phi-4-mini-instruct" or "pytorch/Phi-4-mini-instruct-INT4"
+model_id = "pytorch/Phi-4-mini-instruct-INT4"
 quantized_model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", torch_dtype=torch.bfloat16)
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 
@@ -292,7 +292,7 @@ Our int4wo is only optimized for batch size 1, so expect some slowdown with larg
 ## Results (A100 machine)
 | Benchmark (Latency)              |                |                            |
 |----------------------------------|----------------|----------------------------|
-|                                  | Phi-4 mini-Ins | phi4-mini-int4wo-hqq       | 
+|                                  | Phi-4 mini-Ins | phi4-mini-INT4       | 
 | latency (batch_size=1)           | 2.46s          | 2.2s (1.12x speedup)       |
 | serving (num_prompts=1)          | 0.87 req/s     | 1.05 req/s (1.20x speedup) |
 
@@ -322,9 +322,9 @@ Run the benchmarks under `vllm` root folder:
 python benchmarks/benchmark_latency.py --input-len 256 --output-len 256 --model microsoft/Phi-4-mini-instruct --batch-size 1
 ```
 
-### int4wo-hqq
+### INT4
 ```Shell
-VLLM_DISABLE_COMPILE_CACHE=1 python benchmarks/benchmark_latency.py --input-len 256 --output-len 256 --model pytorch/Phi-4-mini-instruct-int4wo-hqq --batch-size 1
+VLLM_DISABLE_COMPILE_CACHE=1 python benchmarks/benchmark_latency.py --input-len 256 --output-len 256 --model pytorch/Phi-4-mini-instruct-INT4 --batch-size 1
 ```
 
 ## benchmark_serving
@@ -354,15 +354,15 @@ Client:
 python benchmarks/benchmark_serving.py --backend vllm --dataset-name sharegpt --tokenizer microsoft/Phi-4-mini-instruct --dataset-path ./ShareGPT_V3_unfiltered_cleaned_split.json --model microsoft/Phi-4-mini-instruct --num-prompts 1
 ```
 
-### int4wo-hqq
+### INT4
 Server:
 ```Shell
-VLLM_DISABLE_COMPILE_CACHE=1 vllm serve pytorch/Phi-4-mini-instruct-int4wo-hqq --tokenizer microsoft/Phi-4-mini-instruct -O3 --pt-load-map-location cuda:0
+VLLM_DISABLE_COMPILE_CACHE=1 vllm serve pytorch/Phi-4-mini-instruct-INT4 --tokenizer microsoft/Phi-4-mini-instruct -O3 --pt-load-map-location cuda:0
 ```
 
 Client:
 ```Shell
-python benchmarks/benchmark_serving.py --backend vllm --dataset-name sharegpt --tokenizer microsoft/Phi-4-mini-instruct --dataset-path ./ShareGPT_V3_unfiltered_cleaned_split.json --model pytorch/Phi-4-mini-instruct-int4wo-hqq --num-prompts 1
+python benchmarks/benchmark_serving.py --backend vllm --dataset-name sharegpt --tokenizer microsoft/Phi-4-mini-instruct --dataset-path ./ShareGPT_V3_unfiltered_cleaned_split.json --model pytorch/Phi-4-mini-instruct-INT4 --num-prompts 1
 ```
 </details>