File size: 5,582 Bytes
8298b71
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
---
license: mit
tags:
- anomaly-segmentation
---

# 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`.