quangnhvn34

Fix: Configure gradient accumulation and chunking collator for stable training

2

## Summary This PR updates the training configuration to enable gradient accumulation. This change stabilizes the training loss and prevents the `ValueError: matrix contains invalid numeric entries` error, which often...

BUG: Missing key(s) in state_dict: "detector.backbone.vision_backbone.trunk.pos_embed"

1

I get error when inference with custom checkpoint. Train my custom data in rf100 with segment return Sam3VideoPredictorMultiGPU( File "/home/qmask_quangnh58/detect/sam3/sam3/model/sam3_video_predictor.py", line 318, in __init__ super().__init__(*model_args, **model_kwargs) File "/home/qmask_quangnh58/detect/sam3/sam3/model/sam3_video_predictor.py", line 43,...

Bug: ValueError: matrix contains invalid numeric entries

12

INFO 2025-12-04 07:54:24,909 train_utils.py: 268: Train Epoch: [70][200/532] | Batch Time: 0.80 (0.44) | Data Time: 0.00 (0.07) | Mem (GB): 16.00 (16.33/19.00) | Time Elapsed: 00d 04h 24m |...