Native-instance trainings, batch 2 (native route)
The native-instance route's batch-2 trainings (run native_batch2_20261007): UNO variants and residual instance heads, one folder per configuration with its config and run manifest.
165 files, 86.11 GiB. Paths are relative to the root of the research code; index.json lists each
file's size and sha256. Each folder keeps the run manifest (command, commit, configuration) and/or the Detectron2
config.yaml of its training.
| Folder | Checkpoints | Run | Command (first attempt) |
|---|---|---|---|
output/native_batch2/_dropped_b2_detail_tap_seed_base |
1 | native_batch2_20261007 | uv run python train_residual.py --config-file configs/cityscapes/semantic-segmentation/swin/native_b2/b2_detail.yaml --num-gpus 1 --run native_batch2_20261007 --resume SEED 1 OUTPUT_DIR output/native_batch2/b2_detail |
output/native_batch2/a1_seeds20/train |
1 | see config.yaml | |
output/native_batch2/b2_fix |
11 | native_batch2_20261007 | uv run python train_residual.py --config-file configs/cityscapes/semantic-segmentation/swin/native_b2/b2_fix.yaml --num-gpus 1 --run native_batch2_20261007 --resume SEED 1 OUTPUT_DIR output/native_batch2/b2_fix |
output/native_batch2/b2_obj |
11 | native_batch2_20261007 | uv run python train_residual.py --config-file configs/cityscapes/semantic-segmentation/swin/native_b2/b2_obj.yaml --num-gpus 1 --run native_batch2_20261007 --resume SEED 1 OUTPUT_DIR output/native_batch2/b2_obj |
output/native_batch2/b2_r2 |
11 | native_batch2_20261007 | uv run python train_residual.py --config-file configs/cityscapes/semantic-segmentation/swin/native_b2/b2_r2.yaml --num-gpus 1 --run native_batch2_20261007 --resume SEED 1 OUTPUT_DIR output/native_batch2/b2_r2 |
output/native_batch2/b2_road |
11 | native_batch2_20261007 | uv run python train_residual.py --config-file configs/cityscapes/semantic-segmentation/swin/native_b2/b2_road.yaml --num-gpus 1 --run native_batch2_20261007 --resume SEED 1 OUTPUT_DIR output/native_batch2/b2_road |
output/native_batch2/b2_sam3obj |
11 | native_batch2_20261007 | uv run python train_residual.py --config-file configs/cityscapes/semantic-segmentation/swin/native_b2/b2_sam3obj.yaml --num-gpus 1 --run native_batch2_20261007 --resume SEED 1 OUTPUT_DIR output/native_batch2/b2_sam3obj |
output/native_batch2/b2_seed |
11 | native_batch2_20261007 | uv run python train_residual.py --config-file configs/cityscapes/semantic-segmentation/swin/native_b2/b2_seed.yaml --num-gpus 1 --run native_batch2_20261007 --resume SEED 1 OUTPUT_DIR output/native_batch2/b2_seed |
output/native_batch2/b2_tap |
11 | native_batch2_20261007 | uv run python train_residual.py --config-file configs/cityscapes/semantic-segmentation/swin/native_b2/b2_tap.yaml --num-gpus 1 --run native_batch2_20261007 --resume SEED 1 OUTPUT_DIR output/native_batch2/b2_tap |
output/native_batch2/cs_lr3x_mp/train |
1 | see config.yaml | |
output/native_batch2/d1_safe_distill/train |
1 | see config.yaml | |
output/native_batch2/dino_fuse/train |
1 | see config.yaml | |
output/native_batch2/dino_fuse_cnx/train |
1 | see config.yaml | |
output/native_batch2/dino_fuse_lr10/train |
1 | see config.yaml | |
output/native_batch2/g1_blindspot/train |
1 | see config.yaml | |
output/native_batch2/h6_void/train |
1 | see config.yaml | |
output/native_batch2/hires_lr3x_mp/train |
1 | see config.yaml | |
output/native_batch2/pan_mv_it12k/train |
1 | see config.yaml | |
output/native_batch2/pan_mv_lr3x/train |
1 | see config.yaml | |
output/native_batch2/refiner |
2 | native_batch2_20261007 | uv run python scripts/oodis/mask_refiner.py --out output/native_batch2/refiner --resume |
output/native_batch2/sam3_lr3x/train |
1 | see config.yaml | |
output/native_batch2/sam3_lr3x_mp/train |
1 | see config.yaml | |
output/native_batch2/sam3_lr3x_mp20/train |
1 | see config.yaml | |
output/native_batch2/sam3_lr3x_mp_it12k/train |
1 | see config.yaml | |
output/native_batch2/sam3_lr3x_mp_road09/train |
1 | see config.yaml | |
output/native_batch2/sam3_lr6x/train |
1 | see config.yaml | |
output/native_batch2/seam_blur1/train |
1 | see config.yaml | |
output/native_batch2/seam_blur2/train |
1 | see config.yaml | |
output/native_batch2/synth_lr3x_mp/train |
1 | see config.yaml | |
output/native_batch2/uno_bimsn/train |
1 | see config.yaml | |
output/native_batch2/uno_bimsn_road/train |
1 | see config.yaml | |
output/native_batch2/uno_pan_cs/train |
1 | see config.yaml | |
output/native_batch2/uno_pan_mv/train |
1 | see config.yaml | |
output/native_batch2/uno_pan_mv_hires/train |
1 | see config.yaml | |
output/native_batch2/uno_pan_mv_mp/train |
1 | see config.yaml | |
output/native_batch2/uno_pan_mv_sam3/train |
1 | see config.yaml | |
output/native_batch2/uno_pan_mv_sam3_mp/train |
1 | see config.yaml | |
output/native_batch2/uno_pan_mv_synth/train |
1 | see config.yaml | |
output/native_batch2/w1_chiptrain/train |
1 | see config.yaml | |
output/native_batch2/w2a_vistasval/train |
1 | see config.yaml | |
output/native_batch2/w6_frozen/train |
1 | see config.yaml | |
output/native_batch2/w6_tuned/train |
1 | see config.yaml | |
output/native_batch2/w7_safe_input/train |
1 | see config.yaml |
Restore into a checkout of the research code: uv run python scripts/ops/backup_weights.py pull --only native-batch2.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support