train_stsb_42_1760637580

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: 0.4216
• 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: 0.03
• 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
0.4747	1.0	1294	0.4854	435696
0.3948	2.0	2588	0.4649	872288
0.3869	3.0	3882	0.4333	1309872
0.4917	4.0	5176	0.4449	1747344
0.3777	5.0	6470	0.4258	2184032
0.3456	6.0	7764	0.4216	2622912
0.536	7.0	9058	0.4239	3059648
0.3693	8.0	10352	0.4315	3496896
0.3353	9.0	11646	0.4438	3934000
0.3025	10.0	12940	0.4474	4369680
0.3997	11.0	14234	0.4730	4807440
0.2902	12.0	15528	0.4890	5243776
0.2457	13.0	16822	0.5415	5681232
0.2338	14.0	18116	0.5830	6118000
0.1754	15.0	19410	0.6711	6554032
0.1299	16.0	20704	0.7878	6989408
0.1662	17.0	21998	0.9459	7425264
0.1772	18.0	23292	1.0427	7861712
0.0643	19.0	24586	1.0606	8297664
0.1862	20.0	25880	1.0602	8733312

Framework versions

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