Upload folder using huggingface_hub

README.md

@@ -60,7 +60,6 @@ This model is designed as a high-performance offline engine for multimodal tasks
 **MOSS-VL-Instruct-0408** adopts a decoupled multimodal design, utilizing a cross-attention mechanism to bridge high-resolution visual encoding with advanced language reasoning. 
 <p align="center">
     <img src="assets/structure.png" alt="MOSS-VL Architecture" width="90%"/>
-    <em>Figure 1: MOSS-VL Core Architecture.</em>
 </p>
 
 
@@ -69,7 +68,6 @@ This model is designed as a high-performance offline engine for multimodal tasks
 At the model-family level, MOSS-VL uses timestamp-aware multimodal prompting for video understanding. This design gives sampled frames explicit temporal anchors, which helps the model reason about order, duration, and event localization more robustly.
 <p align="center">
     <img src="assets/timestamp_input.svg" alt="Timestamped Sequence Input Illustration" width="90%"/>
-    <em>Figure 2: Illustration of the timestamped sequence input pipeline.</em>
 </p>
 
 ## Multimodal RoPE
@@ -77,7 +75,6 @@ At the model-family level, MOSS-VL uses timestamp-aware multimodal prompting for
 MOSS-VL uses multimodal rotary position encoding to align text tokens and visual features in a shared spatial-temporal coordinate system. At a high level, this improves video-text grounding and helps preserve temporal structure during multimodal reasoning.
 <p align="center">
     <img src="assets/3d-rope.png" alt="MOSS-VL mRoPE Architecture Illustration" width="80%"/>
-    <em>Figure 3: 3D-RoPE spatial-temporal alignment.</em>
 </p>
 
 
@@ -86,7 +83,6 @@ MOSS-VL uses multimodal rotary position encoding to align text tokens and visual
 We evaluate **MOSS-VL-Instruct-0408** across several key multimodal benchmarks, focusing on both video and image understanding. 
 <p align="center">
     <img src="assets/MOSS-VL-Benchmark.png" alt="MOSS-VL Benchmark Results" width="100%"/>
-    <em>Figure 4: Performance comparison on mainstream multimodal benchmarks.</em>
 </p>