train_wsc_123_1760351015

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

• Loss: 0.3529
• Num Input Tokens Seen: 490000

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: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Input Tokens Seen
2.8259	0.504	63	2.4234	25504
0.7315	1.008	126	0.5473	49696
1.1616	1.512	189	0.5321	74112
0.4384	2.016	252	0.3678	99136
0.384	2.52	315	0.3612	123904
0.4165	3.024	378	0.4323	148736
0.4076	3.528	441	0.4275	174432
0.3432	4.032	504	0.3646	198656
0.355	4.536	567	0.3839	224032
0.355	5.04	630	0.3732	247424
0.3847	5.5440	693	0.3541	271232
0.4328	6.048	756	0.3728	295728
0.3633	6.552	819	0.3550	320464
0.2882	7.056	882	0.5431	345856
0.3561	7.5600	945	0.3561	371040
0.3555	8.064	1008	0.3574	395216
0.3562	8.568	1071	0.3547	419184
0.3731	9.072	1134	0.3622	444560
0.3406	9.576	1197	0.3529	469104

Framework versions

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