train_wic_1752870509

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

• Loss: 0.3412
• Num Input Tokens Seen: 4213808

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
3.1888	0.5	611	3.0427	210240
0.4678	1.0	1222	0.4648	421528
0.356	1.5	1833	0.3837	632632
0.3658	2.0	2444	0.3662	843368
0.3545	2.5	3055	0.3566	1054024
0.3267	3.0	3666	0.3541	1264408
0.3546	3.5	4277	0.3530	1475000
0.3166	4.0	4888	0.3487	1685768
0.3747	4.5	5499	0.3450	1895752
0.3684	5.0	6110	0.3437	2106968
0.3188	5.5	6721	0.3423	2318136
0.3403	6.0	7332	0.3435	2528648
0.3211	6.5	7943	0.3416	2739720
0.3119	7.0	8554	0.3412	2949592
0.3343	7.5	9165	0.3419	3160056
0.2871	8.0	9776	0.3424	3371056
0.3477	8.5	10387	0.3440	3581616
0.3369	9.0	10998	0.3420	3792672
0.337	9.5	11609	0.3424	4003136
0.3049	10.0	12220	0.3426	4213808

Framework versions

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