train_stsb_101112_1760638038

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: 3.3077
• Num Input Tokens Seen: 8712528

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: 0.001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 101112
• 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
0.7815	1.0	1294	0.6249	434624
0.4025	2.0	2588	0.5137	869056
0.3955	3.0	3882	0.4564	1304160
0.4456	4.0	5176	0.4475	1740416
0.3391	5.0	6470	0.4580	2175568
0.4272	6.0	7764	0.4402	2611168
0.3837	7.0	9058	0.4400	3047200
0.4752	8.0	10352	0.4363	3482720
0.3787	9.0	11646	0.4365	3918416
0.4897	10.0	12940	0.4465	4355072
0.4996	11.0	14234	0.4438	4790336
0.3837	12.0	15528	0.4662	5227040
0.3313	13.0	16822	0.4599	5662848
0.3999	14.0	18116	0.4828	6099600
0.2922	15.0	19410	0.4877	6534256
0.2866	16.0	20704	0.5080	6968992
0.2577	17.0	21998	0.5373	7405040
0.2791	18.0	23292	0.5483	7840784
0.2941	19.0	24586	0.5602	8276160
0.4091	20.0	25880	0.5618	8712528

Framework versions

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