train_siqa_456_1760637828

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: 0.9780
• Num Input Tokens Seen: 53564288

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: 1e-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
0.5466	2.0	13364	0.5509	5356576
0.287	4.0	26728	0.3669	10713440
0.4431	6.0	40092	0.2721	16068320
0.2202	8.0	53456	0.3200	21422464
0.1417	10.0	66820	0.3849	26778080
0.1952	12.0	80184	0.4473	32135328
0.0002	14.0	93548	0.6453	37494592
0.1969	16.0	106912	0.8511	42850528
0.0	18.0	120276	0.9620	48207328
0.0	20.0	133640	0.9780	53564288

Framework versions

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