train_siqa_456_1760637832

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

• Loss: 0.5714
• Num Input Tokens Seen: 60272064

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
0.4459	1.0	7518	0.5798	3015336
0.62	2.0	15036	0.5762	6029736
0.2437	3.0	22554	0.5730	9044064
0.6211	4.0	30072	0.5736	12056056
0.7361	5.0	37590	0.5734	15070152
0.672	6.0	45108	0.5748	18083976
0.5897	7.0	52626	0.5733	21097056
0.6191	8.0	60144	0.5744	24109664
0.5043	9.0	67662	0.5741	27122784
1.4349	10.0	75180	0.5727	30139392
1.1223	11.0	82698	0.5727	33151800
0.4852	12.0	90216	0.5714	36165976
0.0319	13.0	97734	0.5722	39180248
0.3707	14.0	105252	0.5737	42193928
0.7959	15.0	112770	0.5732	45207272
0.5074	16.0	120288	0.5743	48219232
0.7606	17.0	127806	0.5750	51231624
0.9033	18.0	135324	0.5750	54245832
0.403	19.0	142842	0.5750	57258952
0.6253	20.0	150360	0.5750	60272064

Framework versions

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