mehulshankhapal/internvl3-2B-sft-lora-safety-1

InternVL3-2B-sft-lora-1

This model is a fine-tuned version of OpenGVLab/InternVL3-2B-hf on the train_dataset dataset. It achieves the following results on the evaluation set:

• Loss: 0.2036

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: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 4
• optimizer: Use adamw_torch 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: 50
• num_epochs: 5.0

Training results

Training Loss	Epoch	Step	Validation Loss
1.8698	0.6098	50	0.4710
0.3363	1.2195	100	0.2215
0.3861	1.8293	150	0.2152
0.3101	2.4390	200	0.2142
0.4132	3.0488	250	0.2028
0.2737	3.6585	300	0.2066
0.2247	4.2683	350	0.2016
0.2324	4.8780	400	0.2014

Framework versions

• PEFT 0.15.2
• Transformers 4.52.1
• Pytorch 2.2.2+cu121
• Datasets 3.6.0
• Tokenizers 0.21.2