medsiglip-448-ft-crc100k

This model is a fine-tuned version of google/medsiglip-448 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.3957

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 16
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 5
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss
2.0157	0.1696	50	1.3126
1.2948	0.3393	100	1.2470
1.2212	0.5089	150	1.2748
1.23	0.6785	200	1.2548
1.2481	0.8482	250	1.1919
1.2185	1.0170	300	1.1925
1.1155	1.1866	350	1.2026
1.1485	1.3562	400	1.1624
1.1534	1.5259	450	1.1593
1.1397	1.6955	500	1.1433
1.1196	1.8651	550	1.1361
1.0826	2.0339	600	1.1797
1.0308	2.2036	650	1.1511
1.0708	2.3732	700	1.1081
1.0575	2.5428	750	1.1451
1.0171	2.7125	800	1.1297
1.0203	2.8821	850	1.0809
1.0019	3.0509	900	1.0906
0.9815	3.2205	950	1.1364
0.941	3.3902	1000	1.0855
0.9998	3.5598	1050	1.0776
0.9542	3.7294	1100	1.0548
0.9459	3.8991	1150	1.1044
0.9768	4.0679	1200	1.0920
0.9572	4.2375	1250	1.1189
0.9052	4.4071	1300	1.1566
0.8447	4.5768	1350	1.1299
0.9008	4.7464	1400	1.0913
0.8784	4.9160	1450	1.1010
0.86	5.0848	1500	1.1310
0.8258	5.2545	1550	1.1896
0.84	5.4241	1600	1.2823
0.8637	5.5937	1650	1.2589
0.8488	5.7634	1700	1.1698
0.8298	5.9330	1750	1.2574
0.7697	6.1018	1800	1.2800
0.873	6.2714	1850	1.2326
0.7843	6.4411	1900	1.2283
0.8019	6.6107	1950	1.3048
0.7788	6.7803	2000	1.2452
0.7743	6.9500	2050	1.2635
0.7659	7.1187	2100	1.3144
0.8135	7.2884	2150	1.3114
0.7753	7.4580	2200	1.3197
0.7488	7.6277	2250	1.3256
0.7414	7.7973	2300	1.3678
0.7805	7.9669	2350	1.3413
0.7443	8.1357	2400	1.3732
0.7346	8.3053	2450	1.4000
0.7559	8.4750	2500	1.3945
0.7573	8.6446	2550	1.3864
0.7495	8.8142	2600	1.3861
0.6951	8.9839	2650	1.3784
0.7429	9.1527	2700	1.3987
0.7471	9.3223	2750	1.3968
0.7304	9.4919	2800	1.3966
0.778	9.6616	2850	1.3950
0.6908	9.8312	2900	1.3957
0.7074	10.0	2950	1.3957

Framework versions

• Transformers 4.56.1
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.0