Update README.md

README.md

@@ -12,7 +12,7 @@ pipeline_tag: text-generation
 # Ring-lite-linear-preview
 
 <p align="center">
-    <img src="https://huggingface.co/inclusionAI/Ring-lite-linear-preview/blob/main/ant-bailing.png" width="100"/>
+    <img src="https://huggingface.co/inclusionAI/Ring-lite-linear-preview/resolve/main/ant-bailing.png" width="100"/>
 <p>
 
 <p align="center">
@@ -52,12 +52,12 @@ In terms of the evaluation of reasoning ability， Ring-lite-linear-preview achi
 To evaluate the generation throughput, we deploy Ring-lite-linear and the softmax-attention-based Ring-lite based on vLLM on a single NVIDIA A100 GPU. Specifically, the input sequence length is fixed to 1. The end-to-end (E2E) generation time required for generating output sequences of varying lengths is illustrated below. It is shown in the figure that at 32k output length, Ring-lite-linear-preview achieves 2.2× throughput of Ring-lite.
 
 <p align="center">
-    <img src="https://huggingface.co/inclusionAI/Ring-lite-linear-preview/blob/main/throughput.png" width="600"/>
+    <img src="https://huggingface.co/inclusionAI/Ring-lite-linear-preview/resolve/main/throughput.png" width="600"/>
 <p>
 
 Additionally, to illustrate the advantage in inference speed, we present a comparison between Ring-lite-linear-preview and softmax-attention-based Ring-lite under a batch size of 64 and an output length of 16k (60x speedup). It can be observed that the KV cache usage of Ring-lite-linear-preview is nearly 1/6 that of Ring-lite, and the E2E time is reduced by 27.24% compared with Ring-lite.
 <p align="center">
-    <img src="https://huggingface.co/inclusionAI/Ring-lite-linear-preview/blob/main/inference_speed.gif" width="600"/>
+    <img src="https://huggingface.co/inclusionAI/Ring-lite-linear-preview/resolve/main/inference_speed.gif" width="600"/>
 <p>
 
 More details will be reported in our technical report [TBD]