train_stsb_123_1760637694

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the stsb dataset. It achieves the following results on the evaluation set:

• Loss: 0.7601
• Num Input Tokens Seen: 7745736

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: 4
• eval_batch_size: 4
• seed: 123
• 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_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.676	2.0	2300	0.7993	775352
0.7457	4.0	4600	0.7056	1549400
0.4563	6.0	6900	0.6152	2322888
0.3889	8.0	9200	0.6035	3098264
0.3649	10.0	11500	0.5832	3872240
0.5269	12.0	13800	0.6139	4646992
0.3618	14.0	16100	0.6428	5420880
0.3832	16.0	18400	0.7016	6195056
0.3055	18.0	20700	0.7512	6971336
0.3058	20.0	23000	0.7601	7745736

Framework versions

• PEFT 0.17.1
• Transformers 4.51.3
• Pytorch 2.9.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4