AhamadShaik/SegFormer_RESIZE_x.5

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

• Train Loss: 0.0497
• Train Dice Coef: 0.8670
• Train Iou: 0.7679
• Validation Loss: 0.0477
• Validation Dice Coef: 0.8831
• Validation Iou: 0.7923
• Train Lr: 1e-10
• Epoch: 99

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:

• optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': None, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': True, 'is_legacy_optimizer': False, 'learning_rate': 1e-10, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False}
• training_precision: float32

Training results

Train Loss	Train Dice Coef	Train Iou	Validation Loss	Validation Dice Coef	Validation Iou	Train Lr	Epoch
0.2269	0.5814	0.4248	0.1165	0.7019	0.5504	1e-04	0
0.1305	0.6934	0.5423	0.0877	0.7790	0.6433	1e-04	1
0.1116	0.7311	0.5867	0.0729	0.8299	0.7120	1e-04	2
0.0985	0.7624	0.6241	0.0648	0.8555	0.7491	1e-04	3
0.0918	0.7766	0.6431	0.0711	0.8271	0.7098	1e-04	4
0.0869	0.7877	0.6566	0.0607	0.8552	0.7492	1e-04	5
0.0818	0.7993	0.6722	0.0555	0.8665	0.7662	1e-04	6
0.0753	0.8136	0.6906	0.0544	0.8701	0.7719	1e-04	7
0.0719	0.8216	0.7016	0.0530	0.8725	0.7754	1e-04	8
0.0715	0.8221	0.7027	0.0588	0.8610	0.7579	1e-04	9
0.0673	0.8304	0.7139	0.0502	0.8766	0.7820	1e-04	10
0.0634	0.8388	0.7260	0.0520	0.8757	0.7806	1e-04	11
0.0617	0.8435	0.7328	0.0513	0.8776	0.7831	1e-04	12
0.0731	0.8230	0.7046	0.0540	0.8722	0.7752	1e-04	13
0.0612	0.8439	0.7335	0.0523	0.8749	0.7793	1e-04	14
0.0568	0.8534	0.7473	0.0537	0.8779	0.7842	1e-04	15
0.0549	0.8569	0.7529	0.0486	0.8817	0.7903	5e-06	16
0.0526	0.8607	0.7584	0.0470	0.8849	0.7953	5e-06	17
0.0516	0.8641	0.7633	0.0478	0.8844	0.7946	5e-06	18
0.0523	0.8625	0.7610	0.0483	0.8817	0.7901	5e-06	19
0.0507	0.8662	0.7661	0.0475	0.8842	0.7941	5e-06	20
0.0504	0.8664	0.7665	0.0477	0.8832	0.7924	5e-06	21
0.0504	0.8674	0.7682	0.0474	0.8833	0.7926	5e-06	22
0.0501	0.8655	0.7657	0.0475	0.8833	0.7926	2.5e-07	23
0.0498	0.8677	0.7687	0.0471	0.8845	0.7944	2.5e-07	24
0.0504	0.8665	0.7672	0.0470	0.8846	0.7946	2.5e-07	25
0.0502	0.8677	0.7686	0.0472	0.8844	0.7943	2.5e-07	26
0.0502	0.8662	0.7667	0.0477	0.8833	0.7925	2.5e-07	27
0.0507	0.8667	0.7670	0.0462	0.8853	0.7957	1.25e-08	28
0.0495	0.8685	0.7701	0.0475	0.8841	0.7937	1.25e-08	29
0.0503	0.8669	0.7676	0.0472	0.8840	0.7936	1.25e-08	30
0.0495	0.8689	0.7704	0.0471	0.8854	0.7959	1.25e-08	31
0.0496	0.8681	0.7693	0.0474	0.8844	0.7942	1.25e-08	32
0.0502	0.8665	0.7667	0.0480	0.8823	0.7912	1.25e-08	33
0.0499	0.8663	0.7668	0.0467	0.8852	0.7955	6.25e-10	34
0.0498	0.8668	0.7676	0.0471	0.8844	0.7943	6.25e-10	35
0.0505	0.8653	0.7653	0.0480	0.8821	0.7908	6.25e-10	36
0.0497	0.8687	0.7702	0.0471	0.8847	0.7947	6.25e-10	37
0.0506	0.8660	0.7662	0.0476	0.8838	0.7935	6.25e-10	38
0.0499	0.8678	0.7688	0.0473	0.8849	0.7951	1e-10	39
0.0499	0.8668	0.7676	0.0476	0.8839	0.7935	1e-10	40
0.0500	0.8672	0.7679	0.0478	0.8829	0.7921	1e-10	41
0.0500	0.8670	0.7677	0.0468	0.8845	0.7944	1e-10	42
0.0502	0.8668	0.7673	0.0474	0.8837	0.7932	1e-10	43
0.0500	0.8666	0.7671	0.0476	0.8832	0.7926	1e-10	44
0.0495	0.8682	0.7695	0.0474	0.8839	0.7935	1e-10	45
0.0495	0.8680	0.7690	0.0474	0.8842	0.7938	1e-10	46
0.0502	0.8666	0.7671	0.0474	0.8840	0.7937	1e-10	47
0.0501	0.8668	0.7673	0.0473	0.8840	0.7936	1e-10	48
0.0498	0.8676	0.7686	0.0470	0.8842	0.7939	1e-10	49
0.0495	0.8677	0.7690	0.0477	0.8831	0.7924	1e-10	50
0.0496	0.8694	0.7713	0.0471	0.8846	0.7945	1e-10	51
0.0496	0.8686	0.7699	0.0467	0.8851	0.7953	1e-10	52
0.0495	0.8688	0.7701	0.0469	0.8848	0.7949	1e-10	53
0.0497	0.8677	0.7686	0.0468	0.8848	0.7950	1e-10	54
0.0492	0.8689	0.7704	0.0473	0.8845	0.7944	1e-10	55
0.0498	0.8678	0.7687	0.0473	0.8837	0.7932	1e-10	56
0.0502	0.8668	0.7672	0.0471	0.8838	0.7934	1e-10	57
0.0497	0.8670	0.7676	0.0469	0.8840	0.7936	1e-10	58
0.0500	0.8680	0.7690	0.0473	0.8837	0.7933	1e-10	59
0.0497	0.8681	0.7692	0.0467	0.8840	0.7937	1e-10	60
0.0496	0.8685	0.7694	0.0474	0.8844	0.7944	1e-10	61
0.0506	0.8659	0.7660	0.0474	0.8838	0.7933	1e-10	62
0.0496	0.8677	0.7689	0.0472	0.8850	0.7953	1e-10	63
0.0498	0.8669	0.7675	0.0468	0.8836	0.7930	1e-10	64
0.0498	0.8675	0.7684	0.0471	0.8843	0.7942	1e-10	65
0.0499	0.8680	0.7691	0.0472	0.8842	0.7941	1e-10	66
0.0499	0.8677	0.7688	0.0474	0.8835	0.7928	1e-10	67
0.0501	0.8655	0.7656	0.0466	0.8855	0.7960	1e-10	68
0.0499	0.8673	0.7682	0.0480	0.8825	0.7913	1e-10	69
0.0494	0.8682	0.7698	0.0470	0.8851	0.7955	1e-10	70
0.0499	0.8676	0.7685	0.0475	0.8837	0.7932	1e-10	71
0.0500	0.8672	0.7681	0.0467	0.8855	0.7960	1e-10	72
0.0502	0.8662	0.7664	0.0473	0.8829	0.7919	1e-10	73
0.0498	0.8670	0.7679	0.0474	0.8846	0.7947	1e-10	74
0.0501	0.8665	0.7671	0.0480	0.8827	0.7916	1e-10	75
0.0493	0.8677	0.7689	0.0473	0.8836	0.7930	1e-10	76
0.0496	0.8678	0.7687	0.0474	0.8843	0.7942	1e-10	77
0.0495	0.8679	0.7689	0.0472	0.8844	0.7943	1e-10	78
0.0496	0.8679	0.7690	0.0470	0.8846	0.7945	1e-10	79
0.0501	0.8673	0.7683	0.0473	0.8836	0.7931	1e-10	80
0.0497	0.8679	0.7691	0.0471	0.8839	0.7936	1e-10	81
0.0496	0.8681	0.7693	0.0475	0.8836	0.7931	1e-10	82
0.0495	0.8689	0.7703	0.0474	0.8836	0.7930	1e-10	83
0.0496	0.8685	0.7697	0.0470	0.8845	0.7945	1e-10	84
0.0504	0.8665	0.7669	0.0477	0.8833	0.7926	1e-10	85
0.0496	0.8677	0.7690	0.0478	0.8830	0.7921	1e-10	86
0.0493	0.8682	0.7694	0.0470	0.8837	0.7931	1e-10	87
0.0495	0.8677	0.7688	0.0475	0.8835	0.7929	1e-10	88
0.0499	0.8668	0.7673	0.0471	0.8844	0.7942	1e-10	89
0.0495	0.8682	0.7694	0.0476	0.8836	0.7930	1e-10	90
0.0499	0.8672	0.7679	0.0475	0.8835	0.7929	1e-10	91
0.0496	0.8676	0.7685	0.0478	0.8831	0.7923	1e-10	92
0.0500	0.8677	0.7686	0.0475	0.8838	0.7934	1e-10	93
0.0495	0.8677	0.7686	0.0471	0.8837	0.7931	1e-10	94
0.0494	0.8680	0.7691	0.0473	0.8835	0.7930	1e-10	95
0.0500	0.8656	0.7659	0.0465	0.8848	0.7950	1e-10	96
0.0494	0.8678	0.7690	0.0477	0.8821	0.7907	1e-10	97
0.0498	0.8681	0.7691	0.0475	0.8843	0.7942	1e-10	98
0.0497	0.8670	0.7679	0.0477	0.8831	0.7923	1e-10	99

Framework versions

• Transformers 4.27.1
• TensorFlow 2.11.0
• Datasets 2.10.1
• Tokenizers 0.13.2