Update README.md

README.md

@@ -6,11 +6,9 @@ base_model:
 
 # What's New
 
-Quantized to FP8 using LLM-compressor with similar recipe from [HunyuanImage-3.0-Instruct-Distil-INT8-v2](https://huggingface.co/EricRollei/HunyuanImage-3.0-Instruct-Distil-INT8-v2)
+This repo contains an FP8 quantized HunyuanImage-3.0-Instruct-Distil using LLM-compressor with a similar recipe from [HunyuanImage-3.0-Instruct-Distil-INT8-v2](https://huggingface.co/EricRollei/HunyuanImage-3.0-Instruct-Distil-INT8-v2). Model codes are patched to enable inference with FlashInfer CUTLASS FP8 MoE kernel. Tested on a DGX Spark Founder Edition with FlashInfer==0.6.8 and a modified [Comfy_HunyuanImage3](https://github.com/EricRollei/Comfy_HunyuanImage3) node in ComfyUI. Inference time with cot_recaption reduced from ~1400s to ~200s. Patched ComfyUI nodes will be uploaded soon.
 
-Enabling quantized inference with FlashInfer that fits in a single DGX Spark. Tested on a DGX Spark Founder Edition with FlashInfer==0.6.8 and a modified [Comfy_HunyuanImage3](https://github.com/EricRollei/Comfy_HunyuanImage3) node in ComfyUI.
-
-Inference time with cot_recaption reduced from ~1400s to ~200s on DGX Spark.
+Feel free to open an issue if you encounter any problem when trying to use it.
 
 # Original README