Junhyeok Kim
Junhyeok Kim
Is there a plan for using vicuna delta v1.1? Or is it possible to directly train with v1.1 on my own?
line 222 in label_smoothed_cross_entropy.py updated as the following: From ` net_output[0].masked_fill_(~constraint_masks, -math.inf)` To ` net_output[0] = net_output[0].masked_fill(~constraint_masks, -math.inf)` (https://github.com/OFA-Sys/OFA/pull/341/commits/6a750999afff254623e7df95279776a460a20a50) With the latest code, I got an error typeError: 'tuple' object...
Hi, when I run demo under slurm environment with the python command, this will be ran via sbatch command: ``` python demo/top_down_video_demo_with_mmdet.py \ mmdet_configs/yolox/yolox_tiny_8x8_300e_coco.py \ pretrained_model/yolox_tiny_8x8_300e_coco_20211124_171234-b4047906.pth \ configs/body/2d_kpt_sview_rgb_img/topdown_heatmap/coco/ViTPose_base_coco_256x192.py \ pretrained_model/vitpose_base_coco_aic_mpii.pth...
Hi, thank you for sharing a nice codebase Is there a way to get COCO wholebody keypoints in this repo like original ViTPose?
Hi, thank you for sharing the nice work! I just read the CoTracker3 paper and I wondered how fast CoTracker3 compared to Cotracker1 or 2. What I only know is...