train_stsb_42_1760637583

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: 4.6624
• Num Input Tokens Seen: 8733312

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: 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_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
5.1311	1.0	1294	4.8549	435696
4.533	2.0	2588	4.7069	872288
5.05	3.0	3882	4.6819	1309872
4.33	4.0	5176	4.6632	1747344
4.3298	5.0	6470	4.6738	2184032
4.4377	6.0	7764	4.6624	2622912
4.7251	7.0	9058	4.6698	3059648
4.8367	8.0	10352	4.6727	3496896
4.8132	9.0	11646	4.6701	3934000
4.6233	10.0	12940	4.6722	4369680
4.7196	11.0	14234	4.6824	4807440
4.7721	12.0	15528	4.6716	5243776
4.719	13.0	16822	4.6645	5681232
4.6732	14.0	18116	4.6693	6118000
4.5954	15.0	19410	4.6714	6554032
4.6003	16.0	20704	4.6758	6989408
4.4188	17.0	21998	4.6758	7425264
4.6091	18.0	23292	4.6639	7861712
4.4443	19.0	24586	4.6639	8297664
4.7556	20.0	25880	4.6639	8733312

Framework versions

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