Update README.md

README.md

@@ -16,7 +16,7 @@ base_model:
 ---
 
 [Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B) quantized with [torchao](https://huggingface.co/docs/transformers/main/en/quantization/torchao) int4 weight only quantization, using [hqq](https://mobiusml.github.io/hqq_blog/) algorithm for improved accuracy, by PyTorch team.
-(UPDATE) Use it directly or serve using [vLLM](https://docs.vllm.ai/en/latest/) for 67% VRAM reduction and 12-20% speedup on A100 GPUs.
+(UPDATE) Use it directly or serve using [vLLM](https://docs.vllm.ai/en/latest/) for 67% VRAM reduction and 20%+ speedup on A100 GPUs.
 
 # Inference with vLLM
 Install vllm nightly and torchao nightly to get some recent changes:
@@ -284,8 +284,8 @@ Our int4wo is only optimized for batch size 1, so expect some slowdown with larg
 | Benchmark (Latency)              |                |                          |
 |----------------------------------|----------------|--------------------------|
 |                                  | Qwen3-8B       | Qwen3-8B-int4wo-hqq      | 
-| latency (batch_size=1)           | TODOs          | TODOs (TODO% speedup)       |
-| serving (num_prompts=1)          | TODO req/s     | TODO req/s (20% speedup) |
+| latency (batch_size=1)           | 3.52s          | 2.84s (24% speedup)       |
+| serving (num_prompts=1)          | 0.64 req/s     | 0.79 req/s (23% speedup) |
 
 Note the result of latency (benchmark_latency) is in seconds, and serving (benchmark_serving) is in number of requests per second.
 Int4 weight only is optimized for batch size 1 and short input and output token length, please stay tuned for models optimized for larger batch sizes or longer token length.