Ray Yang

Another question about BoxInst. In the pairwise term, the the probability of making the same prediction = p_i * p_j + (1 - p_i) * (1 - p_j). However, [lines...

@engrjav @engrjav From my experiments, the default settings are best. However, the second question is still not solved.

> Hi, you may change `nms=dict(type='soft_nms', iou_threshold=0.5)` to `nms=dict(type='nms', iou_threshold=0.65)` when doing multi-scale testing, and then see the test results. From my experience, `soft_nms` only brings 0.1 mAP gain when...

> It's a bit weird to see the random drop in performance. I never experienced this in my experiments. But I saw the performance drop when I used `soft-nms` in...

Yeah, that's right. Weights [1, 0] made the ensembled model improve 0.03 mAP than the single model, 0.51 to 0.54 mAP ,but weights [1, 1] only obtained 0.002 mAP. Maybe...

Oh, thanks for your reply!   I will try it.  ---Original--- From: ***@***.***> Date: Mon, Oct 11, 2021 21:51 PM To: ***@***.***>; Cc: ***@***.******@***.***>; Subject: Re: [ZFTurbo/Weighted-Boxes-Fusion] Bbox is 'Nan' when...

> The syntax warning are puzzeling me a bit, I don't get them. Can you tell me your Python version (I use 3.7), Windows or Linux? Latest master version? Thanks...

The ```token_embedding``` is ```nn.Embedding```, so it will cause a type error.

any updates?

您好，请问只使用 SD 的 loss 训练 text encoder 吗？看到 https://github.com/IDEA-CCNL/Fengshenbang-LM/blob/main/fengshen/examples/finetune_taiyi_stable_diffusion/finetune.sh 里面没有 freeze 参数, 那应该 unet 和 text encoder 都没有 freeze；https://github.com/IDEA-CCNL/Fengshenbang-LM/blob/main/fengshen/examples/finetune_taiyi_stable_diffusion/finetune_taiyi_stable_diffusion_example.ipynb 里面 'freeze_text_encoder': False, 'freeze_unet': False