seq2seq_imlla

This model is a fine-tuned version of on the iva_mt_wslot dataset. It achieves the following results on the evaluation set:

• Loss: 6.0658
• Bleu: 0.0042
• Gen Len: 5.8248

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: 4
• total_train_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
7.441	0.9992	636	7.0735	0.0	5.3206
6.5312	2.0	1273	6.4544	0.0175	9.5238
6.0704	2.9992	1909	6.2110	0.0007	4.9967
5.8907	4.0	2546	6.1000	0.0055	6.944
5.7606	4.9961	3180	6.0658	0.0042	5.8248

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.0+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3