segformer-b0-finetuned-segments-toolwear

This model is a fine-tuned version of nvidia/mit-b0 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0491
• Mean Iou: 0.3531
• Mean Accuracy: 0.7062
• Overall Accuracy: 0.7062
• Accuracy Unlabeled: nan
• Accuracy Mass: 0.7062
• Iou Unlabeled: 0.0
• Iou Mass: 0.7062

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 45

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Mass	Iou Unlabeled	Iou Mass
0.3512	1.25	20	0.3893	0.0773	0.1545	0.1545	nan	0.1545	0.0	0.1545
0.2286	2.5	40	0.2047	0.1937	0.3874	0.3874	nan	0.3874	0.0	0.3874
0.1657	3.75	60	0.1423	0.2491	0.4982	0.4982	nan	0.4982	0.0	0.4982
0.1581	5.0	80	0.1117	0.2649	0.5299	0.5299	nan	0.5299	0.0	0.5299
0.1063	6.25	100	0.0943	0.3327	0.6653	0.6653	nan	0.6653	0.0	0.6653
0.0829	7.5	120	0.0782	0.2983	0.5966	0.5966	nan	0.5966	0.0	0.5966
0.0808	8.75	140	0.0740	0.3257	0.6515	0.6515	nan	0.6515	0.0	0.6515
0.0694	10.0	160	0.0725	0.3503	0.7005	0.7005	nan	0.7005	0.0	0.7005
0.0589	11.25	180	0.0663	0.2629	0.5259	0.5259	nan	0.5259	0.0	0.5259
0.0473	12.5	200	0.0604	0.3685	0.7369	0.7369	nan	0.7369	0.0	0.7369
0.0433	13.75	220	0.0569	0.3055	0.6109	0.6109	nan	0.6109	0.0	0.6109
0.0511	15.0	240	0.0546	0.3572	0.7145	0.7145	nan	0.7145	0.0	0.7145
0.04	16.25	260	0.0536	0.3234	0.6467	0.6467	nan	0.6467	0.0	0.6467
0.0365	17.5	280	0.0555	0.3086	0.6171	0.6171	nan	0.6171	0.0	0.6171
0.0314	18.75	300	0.0505	0.3595	0.7191	0.7191	nan	0.7191	0.0	0.7191
0.0295	20.0	320	0.0536	0.3079	0.6159	0.6159	nan	0.6159	0.0	0.6159
0.0337	21.25	340	0.0490	0.3446	0.6891	0.6891	nan	0.6891	0.0	0.6891
0.0325	22.5	360	0.0489	0.3946	0.7891	0.7891	nan	0.7891	0.0	0.7891
0.0314	23.75	380	0.0514	0.3184	0.6368	0.6368	nan	0.6368	0.0	0.6368
0.0267	25.0	400	0.0485	0.3572	0.7144	0.7144	nan	0.7144	0.0	0.7144
0.0321	26.25	420	0.0490	0.3787	0.7573	0.7573	nan	0.7573	0.0	0.7573
0.025	27.5	440	0.0474	0.3615	0.7230	0.7230	nan	0.7230	0.0	0.7230
0.0225	28.75	460	0.0472	0.3660	0.7319	0.7319	nan	0.7319	0.0	0.7319
0.0247	30.0	480	0.0502	0.3488	0.6976	0.6976	nan	0.6976	0.0	0.6976
0.0216	31.25	500	0.0483	0.3536	0.7072	0.7072	nan	0.7072	0.0	0.7072
0.0195	32.5	520	0.0508	0.3289	0.6578	0.6578	nan	0.6578	0.0	0.6578
0.0259	33.75	540	0.0496	0.3846	0.7692	0.7692	nan	0.7692	0.0	0.7692
0.0242	35.0	560	0.0487	0.3464	0.6928	0.6928	nan	0.6928	0.0	0.6928
0.0217	36.25	580	0.0503	0.3325	0.6650	0.6650	nan	0.6650	0.0	0.6650
0.0204	37.5	600	0.0502	0.3429	0.6858	0.6858	nan	0.6858	0.0	0.6858
0.0204	38.75	620	0.0507	0.3457	0.6913	0.6913	nan	0.6913	0.0	0.6913
0.0191	40.0	640	0.0494	0.3494	0.6988	0.6988	nan	0.6988	0.0	0.6988
0.0204	41.25	660	0.0503	0.3426	0.6852	0.6852	nan	0.6852	0.0	0.6852
0.019	42.5	680	0.0485	0.3616	0.7232	0.7232	nan	0.7232	0.0	0.7232
0.0198	43.75	700	0.0494	0.3504	0.7008	0.7008	nan	0.7008	0.0	0.7008
0.0212	45.0	720	0.0491	0.3531	0.7062	0.7062	nan	0.7062	0.0	0.7062

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2