super.single430

> 请问是否可以将vpm和resampler 这2个模型转为tensorrt来加速推理呢? 这两个模块的耗时貌似并不是大头

> Hi, I'm getting good results with my custom data, but I don't know how to edit dynamic objects, like translating or rotating a vehicle, etc. Can you provide some...

> it's common issue, if you try enough bigger model, the issue is higher 首先非常感谢工作，非常出色，但是如果一味的追求刷榜，考虑不全面，不去关注这些问题，那也是不够的，起码我用qwen3-vl，暂时没有这些问题。

官方demo ``` # multi-image multi-round conversation, separate images (多图多轮对话，独立图像) pixel_values1 = load_image('./examples/image1.jpg', max_num=12).to(torch.bfloat16).cuda() pixel_values2 = load_image('./examples/image2.jpg', max_num=12).to(torch.bfloat16).cuda() pixel_values = torch.cat((pixel_values1, pixel_values2), dim=0) num_patches_list = [pixel_values1.size(0), pixel_values2.size(0)] question = 'Image-1: \nImage-2:...

有参数他也不会生成`tool_calls`标签，`chunk.choices[0].delta.tool_calls`始终为空。我测试了上面输出都是包裹在``中； ``` content: {"name": "get_current_weather", "arguments": {"location": "杭州"}} ``` 解析：`json.loads(response_str.split('\n')[1].split('\n')[0])`