train_siqa_123_1760637715

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: 0.5491
• Num Input Tokens Seen: 60276872

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: 123
• 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
0.5543	1.0	7518	0.5503	3014896
0.5493	2.0	15036	0.5501	6029360
0.5515	3.0	22554	0.5504	9042368
0.5546	4.0	30072	0.5503	12055456
0.5403	5.0	37590	0.5498	15068512
0.5456	6.0	45108	0.5491	18081672
0.544	7.0	52626	0.5497	21095960
0.5473	8.0	60144	0.5500	24109392
0.5472	9.0	67662	0.5500	27122856
0.563	10.0	75180	0.5497	30137256
0.5541	11.0	82698	0.5501	33151024
0.5524	12.0	90216	0.5495	36165000
0.5531	13.0	97734	0.5496	39178496
0.5466	14.0	105252	0.5494	42193184
0.5457	15.0	112770	0.5497	45206576
0.552	16.0	120288	0.5497	48220600
0.5534	17.0	127806	0.5497	51235344
0.5487	18.0	135324	0.5500	54249648
0.5466	19.0	142842	0.5496	57262600
0.5487	20.0	150360	0.5496	60276872

Framework versions

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