IamMegatron

Add swin-transformer to ZOO

5

Hello, as we know, the basic ability of `swin-transformer` is better than `vit`. Can you provide `swin-transformer` in `oml/models` (including pre-train weight) ? upd: [Contribution instruction](https://github.com/OML-Team/open-metric-learning/blob/main/CONTRIBUTING.md#contributing-to-models-zoo).

我发现一个问题哈：检测出一个物体目标框后，随着时间这个框慢慢变的越来越小，这是怎么回事呢？

Sorry to trouble you. The images in `README.md` all are broken. Can you update them?

### Reminder - [x] I have read the above rules and searched the existing issues. ### Description 如题，使用vllm_infer.py进行推理时，怎么返回输出token的prob呢？`results = LLM(**engine_args).generate(inputs, sampling_params, lora_request=lora_request)` 这个函数应该怎么修改呢？ 求助，急急急~~~ ### Pull Request _No response_

How to train on multi gpus?

4

### 🐛 Describe the bug Hello, I follow [docs ](the https://github.com/allenai/olmocr/tree/main/olmocr/train) and run ``` python -m olmocr.train.train --config olmocr/train/configs/qwen25_vl_olmocrv3_1epoch.yaml ``` It could train on the [olmOCR-mix-0225](https://huggingface.co/datasets/allenai/olmOCR-mix-0225) normally on single gpu....