kosm-checkpoint

This model is a fine-tuned version of microsoft/kosmos-2-patch14-224 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0340

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: 1e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.0497	200	0.0700
0.0802	0.0993	400	0.0581
0.0676	0.1490	600	0.0496
0.0584	0.1986	800	0.0450
0.0582	0.2483	1000	0.0481
0.0582	0.2979	1200	0.0486
0.0572	0.3476	1400	0.0445
0.0537	0.3972	1600	0.0463
0.0504	0.4469	1800	0.0421
0.0473	0.4965	2000	0.0402
0.0473	0.5462	2200	0.0423
0.046	0.5958	2400	0.0394
0.0448	0.6455	2600	0.0369
0.0423	0.6951	2800	0.0378
0.0403	0.7448	3000	0.0360
0.0403	0.7944	3200	0.0364
0.0392	0.8441	3400	0.0352
0.0388	0.8937	3600	0.0347
0.0375	0.9434	3800	0.0343
0.037	0.9930	4000	0.0345
0.037	1.0427	4200	0.0355
0.03	1.0924	4400	0.0338
0.0283	1.1420	4600	0.0349
0.0281	1.1917	4800	0.0347
0.0288	1.2413	5000	0.0322
0.0288	1.2910	5200	0.0331
0.0279	1.3406	5400	0.0335
0.0272	1.3903	5600	0.0322
0.0275	1.4399	5800	0.0338
0.0271	1.4896	6000	0.0324
0.0271	1.5392	6200	0.0324
0.0263	1.5889	6400	0.0320
0.0262	1.6385	6600	0.0319
0.0264	1.6882	6800	0.0317
0.0256	1.7378	7000	0.0322
0.0256	1.7875	7200	0.0320
0.0255	1.8371	7400	0.0316
0.0242	1.8868	7600	0.0327
0.0262	1.9364	7800	0.0307
0.0252	1.9861	8000	0.0304
0.0252	2.0357	8200	0.0343
0.0173	2.0854	8400	0.0373
0.0148	2.1351	8600	0.0345
0.015	2.1847	8800	0.0347
0.0148	2.2344	9000	0.0347
0.0148	2.2840	9200	0.0354
0.0132	2.3337	9400	0.0351
0.0136	2.3833	9600	0.0362
0.0132	2.4330	9800	0.0360
0.0138	2.4826	10000	0.0352
0.0138	2.5323	10200	0.0359
0.0138	2.5819	10400	0.0348
0.0132	2.6316	10600	0.0348
0.0129	2.6812	10800	0.0337
0.0134	2.7309	11000	0.0354
0.0134	2.7805	11200	0.0350
0.0132	2.8302	11400	0.0351
0.0128	2.8798	11600	0.0350
0.013	2.9295	11800	0.0339
0.012	2.9791	12000	0.0340

Framework versions

• Transformers 4.42.4
• Pytorch 2.1.2+cu121
• Datasets 2.15.0
• Tokenizers 0.19.1