qwen2-2b-instruct-trl-sft-ChartQA

This model is a fine-tuned version of Qwen/Qwen2-VL-2B-Instruct on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0574

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: 0.0002
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_ratio: 0.03
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
4.1946	0.1546	10	3.2365
2.5247	0.3092	20	1.7800
1.1177	0.4638	30	0.5485
0.4772	0.6184	40	0.4416
0.4286	0.7729	50	0.4205
0.398	0.9275	60	0.3643
0.3027	1.0821	70	0.2035
0.1021	1.2367	80	0.0645
0.0591	1.3913	90	0.0597
0.0592	1.5459	100	0.0581
0.0575	1.7005	110	0.0595
0.0608	1.8551	120	0.0586
0.0583	2.0097	130	0.0588
0.0599	2.1643	140	0.0570
0.0569	2.3188	150	0.0579
0.0577	2.4734	160	0.0574
0.0536	2.6280	170	0.0565
0.0567	2.7826	180	0.0583
0.0611	2.9372	190	0.0574

Framework versions

• PEFT 0.11.1
• Transformers 4.45.2
• Pytorch 2.3.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0