Hangbo Bao

We use a specified version of mmseg lib(mmcv-full==1.3.0 & mmsegmentation==0.11.0) for reproduction. Please follow the [introduction](https://github.com/microsoft/unilm/tree/master/beit/semantic_segmentation#getting-started) for installing the required packages.

> > We use a specified version of mmseg lib(mmcv-full==1.3.0 & mmsegmentation==0.11.0) for reproduction. Please follow the [introduction](https://github.com/microsoft/unilm/tree/master/beit/semantic_segmentation#getting-started) for installing the required packages. > > But this version of mmsegmentation...

Hi everyone, [VLMo code & models](https://github.com/microsoft/unilm/tree/master/vlmo) has been released. Thanks for your interest!

Hi everyone, [VLMo code & models](https://github.com/microsoft/unilm/tree/master/vlmo) has been released. Thanks for your interest!

The code and pre-trained models have been [merged](https://github.com/microsoft/unilm/pull/1024). You can access models and code via [aka.ms/beit3](https://aka.ms/beit3).

Hi @amandaluof , we use the following code to preprocess the class names: ```python from nltk.corpus import wordnet as wn syn = wn.synset_from_pos_and_offset('n', wordnet_id) class_name = ", ".join(_.name() for _...

The code and pre-trained models have been [merged](https://github.com/microsoft/unilm/pull/1024). You can access models and code via [aka.ms/beit3](https://aka.ms/beit3).

Hi @jsrdcht , could you please provide additional information, such as the specific commands being run?

`--auto_resume` will find the latest checkpoint, more details can be found at this [url](https://github.com/microsoft/unilm/blob/master/beit/utils.py#L477)

interpolate the position embedding, more details can be found at [here](https://github.com/microsoft/unilm/blob/master/beit2/modeling_finetune.py#L383) and [here](https://github.com/microsoft/unilm/blob/master/beit2/run_class_finetuning.py#L396)