gm-lora-bfloat16-idefics2-8b-xrayvqa-finetuned-medir2

This model is a fine-tuned version of HuggingFaceM4/idefics2-8b on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.6349

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 10
• total_train_batch_size: 80
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
1.1259	0.0764	50	1.4468
1.2502	0.1529	100	1.4544
1.2599	0.2293	150	1.4605
1.1477	0.3058	200	1.4844
1.1041	0.3822	250	1.4835
1.0958	0.4586	300	1.4724
1.0975	0.5351	350	1.4800
1.133	0.6115	400	1.4656
1.1785	0.6880	450	1.4458
1.3751	0.7644	500	1.4227
1.3751	0.8409	550	1.4187
1.3983	0.9173	600	1.4158
1.4147	0.9937	650	1.4073
0.9615	1.0702	700	1.4901
0.9026	1.1466	750	1.5204
0.8919	1.2231	800	1.4997
0.917	1.2995	850	1.4994
0.9149	1.3759	900	1.4998
0.9342	1.4524	950	1.4971
0.9363	1.5288	1000	1.5039
0.9087	1.6053	1050	1.4907
0.9272	1.6817	1100	1.4920
0.9195	1.7581	1150	1.4955
0.9488	1.8346	1200	1.4900
0.9209	1.9110	1250	1.4887
0.9463	1.9875	1300	1.4891
0.7123	2.0639	1350	1.6077
0.646	2.1403	1400	1.6182
0.6405	2.2168	1450	1.6390
0.6481	2.2932	1500	1.6198
0.6372	2.3697	1550	1.6340
0.6618	2.4461	1600	1.6311
0.6499	2.5226	1650	1.6277
0.6471	2.5990	1700	1.6344
0.6554	2.6754	1750	1.6303
0.6475	2.7519	1800	1.6333
0.641	2.8283	1850	1.6315
0.6274	2.9048	1900	1.6343
0.6309	2.9812	1950	1.6349

Framework versions

• Transformers 4.41.0.dev0
• Pytorch 2.0.1
• Datasets 2.19.1
• Tokenizers 0.19.1