train_qnli_1756735870

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

• Loss: 0.1076
• Num Input Tokens Seen: 94426336

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: 2
• eval_batch_size: 2
• 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: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
0.1101	0.5	23567	0.0919	4726160
0.0048	1.0	47134	0.0505	9443856
0.1224	1.5	70701	0.0457	14171744
0.0074	2.0	94268	0.0379	18885616
0.0206	2.5	117835	0.0410	23594480
0.0022	3.0	141402	0.0483	28322288
0.0055	3.5	164969	0.0407	33044192
0.1087	4.0	188536	0.0372	37765424
0.1327	4.5	212103	0.0544	42484624
0.0309	5.0	235670	0.0420	47208432
0.0031	5.5	259237	0.0485	51927872
0.0863	6.0	282804	0.0511	56653952
0.0032	6.5	306371	0.0703	61379616
0.0021	7.0	329938	0.0588	66100352
0.0002	7.5	353505	0.0728	70821872
0.0012	8.0	377072	0.0659	75542800
0.114	8.5	400639	0.0899	80264848
0.0	9.0	424206	0.0903	84986304
0.002	9.5	447773	0.1065	89703232
0.0002	10.0	471340	0.1076	94426336

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.8.0+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1