gm-lora-bfloat16-idefics2-8b-xrayvqa-finetuned-mimic-short

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: 1.0730

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: 4

Training results

Training Loss	Epoch	Step	Validation Loss
1.171	0.1119	50	1.2339
1.2051	0.2237	100	1.1744
1.1372	0.3356	150	1.1323
1.1046	0.4474	200	1.1095
1.0673	0.5593	250	1.0877
1.0713	0.6711	300	1.0761
1.0824	0.7830	350	1.0611
1.0358	0.8949	400	1.0511
1.0288	1.0067	450	1.0389
0.9158	1.1186	500	1.0443
0.9199	1.2304	550	1.0382
0.9032	1.3423	600	1.0339
0.8834	1.4541	650	1.0292
0.9048	1.5660	700	1.0271
0.9101	1.6779	750	1.0193
0.8928	1.7897	800	1.0164
0.9032	1.9016	850	1.0124
0.8649	2.0134	900	1.0234
0.7615	2.1253	950	1.0433
0.7588	2.2371	1000	1.0366
0.7759	2.3490	1050	1.0331
0.7696	2.4609	1100	1.0349
0.7587	2.5727	1150	1.0324
0.7532	2.6846	1200	1.0309
0.7702	2.7964	1250	1.0287
0.7648	2.9083	1300	1.0275
0.7452	3.0201	1350	1.0529
0.6471	3.1320	1400	1.0683
0.665	3.2438	1450	1.0727
0.6563	3.3557	1500	1.0713
0.6499	3.4676	1550	1.0721
0.6538	3.5794	1600	1.0741
0.6437	3.6913	1650	1.0740
0.6486	3.8031	1700	1.0734
0.66	3.9150	1750	1.0730

Framework versions

• Transformers 4.41.0.dev0
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1