Llama1B_stsb

This model is a fine-tuned version of meta-llama/Llama-3.2-1B on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4672
• Pearson: 0.8971
• Spearman: 0.8984
• Combined Score: 0.8978

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: 3e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Pearson	Spearman	Combined Score
12.6109	1.0	180	0.8516	0.8310	0.8346	0.8328
1.0811	2.0	360	0.6817	0.8762	0.8756	0.8759
0.6641	3.0	540	0.6724	0.8695	0.8740	0.8718
0.4579	4.0	720	0.5094	0.8857	0.8859	0.8858
0.3244	5.0	900	0.5906	0.8908	0.8897	0.8903
0.2715	6.0	1080	0.5000	0.8867	0.8881	0.8874
0.1594	7.0	1260	0.4783	0.8941	0.8943	0.8942
0.102	8.0	1440	0.4699	0.8972	0.8975	0.8974
0.068	9.0	1620	0.4588	0.8973	0.8976	0.8975
0.0351	10.0	1800	0.4672	0.8971	0.8984	0.8978

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0