train_qnli_1752826676

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: 0.0972
• Num Input Tokens Seen: 103607072

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: 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
0.1856	0.5000	11784	0.1695	5193280
0.1548	1.0000	23568	0.1487	10365728
0.1388	1.5001	35352	0.1395	15547488
0.1415	2.0001	47136	0.1304	20725792
0.1585	2.5001	58920	0.1231	25887456
0.1181	3.0001	70704	0.1223	31082368
0.1334	3.5001	82488	0.1142	36266176
0.117	4.0002	94272	0.1107	41440992
0.1282	4.5002	106056	0.1088	46618176
0.1108	5.0002	117840	0.1055	51803520
0.1112	5.5002	129624	0.1035	56978912
0.1045	6.0003	141408	0.1025	62167168
0.0781	6.5003	153192	0.1002	67356288
0.0491	7.0003	164976	0.0992	72532096
0.0818	7.5003	176760	0.0997	77710656
0.0553	8.0003	188544	0.0980	82887904
0.0862	8.5004	200328	0.0976	88066400
0.1003	9.0004	212112	0.0974	93248224
0.1071	9.5004	223896	0.0972	98430752

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.7.1+cu126
• Datasets 3.6.0
• Tokenizers 0.21.1