swin-large-mask2former-finetuned-Mitoliked-Mito-Innertongue2

This model is a fine-tuned version of Dnq2025/swin-large-mask2former-finetuned-ER-Mito-LD8 on the Dnq2025/Mask2former_Finetune dataset. It achieves the following results on the evaluation set:

• Mean Iou: 0.4460
• Loss: 40.5938

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: 5632

Training results

Training Loss	Epoch	Step	Mean Iou	Validation Loss
43.3313	2.0	176	0.3371	35.6391
31.1848	4.0	352	0.3681	34.4077
26.4868	6.0	528	0.3798	35.5335
23.4201	8.0	704	0.4136	34.2507
21.2972	10.0	880	0.4231	34.8946
19.7902	12.0	1056	0.3975	36.7693
18.2954	14.0	1232	0.4123	35.3214
17.1452	16.0	1408	0.4215	35.4186
16.3828	18.0	1584	0.4043	36.7854
15.5539	20.0	1760	0.4371	35.7144
14.9365	22.0	1936	0.4275	36.8812
14.4327	24.0	2112	0.4051	36.8868
13.6736	26.0	2288	0.4119	37.0440
13.2255	28.0	2464	0.4328	37.2634
12.7617	30.0	2640	0.4252	39.2454
12.3143	32.0	2816	0.4372	37.5345
11.9953	34.0	2992	0.4401	37.8913
11.6438	36.0	3168	0.4501	38.7431
11.3325	38.0	3344	0.4463	37.0448
11.0669	40.0	3520	0.4441	38.5237
10.9012	42.0	3696	0.4313	40.2165
10.6703	44.0	3872	0.4379	40.0815
10.4866	46.0	4048	0.4505	40.1928
10.3071	48.0	4224	0.4421	38.1996
10.144	50.0	4400	0.4473	39.9193
10.0052	52.0	4576	0.4482	38.6817
9.9162	54.0	4752	0.4481	39.9527
9.7836	56.0	4928	0.4498	39.1961
9.6469	58.0	5104	0.4483	40.5048
9.6146	60.0	5280	0.4471	41.2389
9.5224	62.0	5456	0.4455	40.9301
9.4583	64.0	5632	0.4460	40.8836

Framework versions

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