Kaixin WANG

In some cases, collecting logs from each worker is pretty useful. For example, when training RL agents, wandb allows collecting metrics from each actor and learner. So users do not...

@kailigo I managed to run the code but didn't reach the reported performance (for the first split, only 0.35). Maybe I still miss something. The proposed architecture is not very...

@kailigo Thanks. It will be helpful. I am afraid I don't have time looking into this right now but I will try it again later and share my experience if...

Hi @704051419 , sorry for the late reply. You are right about the n ways situation. `if target_idx_j` will cause underestimation of FP when target image does not contain all...

Hi @shankrmahadevan , the data loading code is quite VOC/COCO specific. If you want to use an external dataset, I suggest you checking out [torchmeta](https://github.com/tristandeleu/pytorch-meta). It has a cleaner abstraction...

Hi @karry-code , the data loading code is quite VOC/COCO specific. If you want to use an external dataset, I suggest you checking out [torchmeta](https://github.com/tristandeleu/pytorch-meta). It has a cleaner abstraction...

Hi @chunbolang , `scribble_dilation` is used to increase the width of the scribbles, so using a large value can produce annotations similar to ones in Figure 6. I checked the...

Hi @lailvlong , I tried resnet50 before, and as far as I can remember, the performance is slightly better (~1%). According to some recent papers I read, the performance gap...

Hi @azharshaikh06 , you can modify configurations in `config.py` accordingly for running N-way experiments. Please note that there is a bug when `n_queries != 1` (see #6). Thank you.

@ritalaezza The model is initialized in agent's initialization. Most `__init__()` methods in `rlpyt` only save the args and real initialization is done in `initialize()` method. For you question, one workaround...