train_stsb_456_1760637813

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: 4.6793
• Num Input Tokens Seen: 8714656

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: 4
• eval_batch_size: 4
• seed: 456
• 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
4.8014	1.0	1294	4.8659	435104
4.5337	2.0	2588	4.7270	870112
4.5882	3.0	3882	4.6800	1305024
4.781	4.0	5176	4.6901	1742048
4.6547	5.0	6470	4.6899	2176672
4.792	6.0	7764	4.6986	2613648
4.7334	7.0	9058	4.6904	3049776
4.8957	8.0	10352	4.6945	3486928
4.7409	9.0	11646	4.6894	3924192
4.7251	10.0	12940	4.6873	4360736
4.7489	11.0	14234	4.6855	4793520
4.8942	12.0	15528	4.6980	5230528
4.9596	13.0	16822	4.6977	5664848
4.4292	14.0	18116	4.6952	6100288
4.3208	15.0	19410	4.6901	6534240
4.922	16.0	20704	4.6876	6969936
4.7687	17.0	21998	4.6793	7405056
4.9178	18.0	23292	4.6793	7842624
4.5854	19.0	24586	4.6793	8279952
4.6684	20.0	25880	4.6793	8714656

Framework versions

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