segformer-b2-detection

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2432
• Mean Iou: 0.3818

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 1000
• training_steps: 40000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou
0.1722	0.025	1000	0.2283	0.3572
0.1192	0.05	2000	0.2526	0.3822
0.1456	0.075	3000	0.2463	0.3827
0.154	0.1	4000	0.2447	0.3632
0.1688	0.125	5000	0.2146	0.3598
0.1725	0.15	6000	0.2148	0.3287
0.1522	0.175	7000	0.1987	0.3776
0.1454	0.2	8000	0.2618	0.3824
0.1631	1.0082	9000	0.3216	0.3516
0.1428	1.0332	10000	0.2662	0.3823
0.1439	1.0582	11000	0.2530	0.3502
0.1148	1.0832	12000	0.2817	0.3844
0.1299	1.1082	13000	0.2481	0.3618
0.155	1.1332	14000	0.2587	0.3718
0.166	1.1582	15000	0.2753	0.3708
0.1505	1.1832	16000	0.2401	0.4050
0.1293	1.2082	17000	0.2345	0.3551
0.1375	2.0164	18000	0.2423	0.3611
0.1721	2.0413	19000	0.2239	0.3950
0.1448	2.0663	20000	0.2273	0.3564
0.1067	2.0913	21000	0.2268	0.3851
0.1497	2.1164	22000	0.2340	0.3524
0.1436	2.1414	23000	0.2465	0.3718
0.1402	2.1663	24000	0.2669	0.3817
0.1224	2.1913	25000	0.2416	0.3895
0.1813	2.2163	26000	0.2375	0.3681
0.1361	3.0245	27000	0.2723	0.3833
0.1506	3.0495	28000	0.2838	0.3764
0.1367	3.0745	29000	0.2709	0.3473
0.1162	3.0995	30000	0.2685	0.3930
0.1803	3.1245	31000	0.2489	0.3886
0.0907	3.1495	32000	0.2487	0.3869
0.1158	3.1745	33000	0.2422	0.3885
0.1236	3.1995	34000	0.2326	0.3923
0.1412	4.0077	35000	0.2295	0.3885
0.1287	4.0327	36000	0.2405	0.3844
0.1525	4.0577	37000	0.2429	0.3857
0.133	4.0827	38000	0.2438	0.3829
0.146	4.1077	39000	0.2381	0.3840
0.1142	4.1327	40000	0.2432	0.3818

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu126
• Datasets 4.4.2
• Tokenizers 0.22.1