VectorSpaceLab / Video-XL

🔥🔥First-ever hour scale video understanding models
Apache License 2.0
183 stars 12 forks source link

Comparison of experimental results on VNBench and LongVideoBench with Qwen-VL #7

Open xesdiny opened 1 month ago

xesdiny commented 1 month ago

They are all based on Qwen2, so why not compare Qwen2-VL on the benchmark to compare the biggest change, the Encoder (patchify embedding -> CLIP)?

shuyansy commented 1 month ago

Thanks for your advice. We will compare Qwen2-VL on more benchmarks in the following reports