train_wic_1754502825

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: 0.2365
• 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
0.3932	0.5	611	0.3330	210240
0.3046	1.0	1222	0.3138	421528
0.2997	1.5	1833	0.2763	632632
0.1182	2.0	2444	0.3277	843368
0.3365	2.5	3055	0.2781	1054024
0.2174	3.0	3666	0.2534	1264408
0.2857	3.5	4277	0.2595	1475000
0.2087	4.0	4888	0.2712	1685768
0.2269	4.5	5499	0.2482	1895752
0.326	5.0	6110	0.2451	2106968
0.385	5.5	6721	0.2522	2318136
0.2193	6.0	7332	0.2365	2528648
0.2994	6.5	7943	0.2465	2739720
0.1564	7.0	8554	0.2551	2949592
0.3183	7.5	9165	0.2515	3160056
0.2118	8.0	9776	0.2540	3371056
0.2072	8.5	10387	0.2524	3581616
0.2059	9.0	10998	0.2518	3792672
0.2168	9.5	11609	0.2549	4003136
0.0975	10.0	12220	0.2532	4213808

Framework versions

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