qg-isg-simp-m-small-ls0.0

This model is a fine-tuned version of bigscience/mt0-small on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5954

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: 0.0003
• train_batch_size: 48
• eval_batch_size: 48
• seed: 42
• 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: 500
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
0.7838	0.3700	10000	0.7116
0.697	0.7401	20000	0.6674
0.648	1.1101	30000	0.6371
0.6128	1.4801	40000	0.6137
0.5889	1.8501	50000	0.6073
0.5639	2.2202	60000	0.5975
0.5953	2.5902	70000	0.5960
0.5654	2.9602	80000	0.5954

Framework versions

• Transformers 4.51.3
• Pytorch 2.5.0a0+e000cf0ad9.nv24.10
• Datasets 3.6.0
• Tokenizers 0.21.1