mask2former-finetuned-ER-Mito-LD7

This model is a fine-tuned version of facebook/mask2former-swin-large-ade-semantic on the Dnq2025/Mask2former_Pretrain dataset. It achieves the following results on the evaluation set:

• Loss: 32.5161

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: 0.0004
• train_batch_size: 4
• eval_batch_size: 4
• seed: 1337
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: polynomial
• training_steps: 6450

Training results

Training Loss	Epoch	Step	Validation Loss
49.96	1.0	129	38.5311
37.9944	2.0	258	33.0716
36.2103	3.0	387	29.7335
29.2814	4.0	516	31.5442
27.3516	5.0	645	28.0345
26.4039	6.0	774	27.9411
24.3475	7.0	903	26.4108
23.4804	8.0	1032	26.8475
22.4815	9.0	1161	25.9447
21.721	10.0	1290	27.1656
20.8443	11.0	1419	33.2659
19.9949	12.0	1548	26.9611
19.893	13.0	1677	26.0445
18.1322	14.0	1806	27.2854
18.1679	15.0	1935	25.4194
17.4814	16.0	2064	25.4006
17.402	17.0	2193	24.8677
17.0285	18.0	2322	25.9922
15.8946	19.0	2451	27.1687
15.8518	20.0	2580	29.3397
15.4202	21.0	2709	25.7427
14.9686	22.0	2838	28.8585
14.7436	23.0	2967	27.9649
15.1461	24.0	3096	27.5371
14.3666	25.0	3225	27.2910
13.9871	26.0	3354	28.4562
13.8003	27.0	3483	27.0616
13.7903	28.0	3612	33.0673
13.2151	29.0	3741	28.1574
13.2489	30.0	3870	27.9714
12.9787	31.0	3999	29.6233
12.8853	32.0	4128	32.2755
12.5442	33.0	4257	30.2798
13.3521	34.0	4386	28.7282
12.609	35.0	4515	27.4472
12.1436	36.0	4644	28.6240
12.0534	37.0	4773	28.3248
12.4731	38.0	4902	31.0330
12.0568	39.0	5031	33.3478
11.7165	40.0	5160	32.6755
11.7194	41.0	5289	32.9583
11.5118	42.0	5418	31.8171
11.2862	43.0	5547	30.4766
11.8368	44.0	5676	30.9541
11.2132	45.0	5805	32.4065
10.606	46.0	5934	31.5392
11.7442	47.0	6063	31.8038
10.7855	48.0	6192	32.5302
11.3661	49.0	6321	32.8178
10.9675	50.0	6450	32.5147

Framework versions

• Transformers 4.50.0.dev0
• Pytorch 2.4.1
• Datasets 3.3.2
• Tokenizers 0.21.0