2603c9d41080a7774fa78ac465645aa6

This model is a fine-tuned version of meta-llama/Llama-3.2-1B on the nyu-mll/glue [qnli] dataset. It achieves the following results on the evaluation set:

• Loss: 2.9518
• Data Size: 1.0
• Epoch Runtime: 432.6197
• Accuracy: 0.7688
• F1 Macro: 0.7680
• Rouge1: 0.7686
• Rouge2: 0.0
• Rougel: 0.7691
• Rougelsum: 0.7686

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: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro	Rouge1	Rouge2	Rougel	Rougelsum
No log	0	0	4.1669	0	7.7608	0.5092	0.5025	0.5089	0.0	0.5086	0.5092
No log	1	3273	2.7455	0.0078	10.7155	0.5625	0.5336	0.5628	0.0	0.5629	0.5623
0.0926	2	6546	2.9561	0.0156	14.7526	0.5029	0.3889	0.5033	0.0	0.5035	0.5028
2.7438	3	9819	2.8563	0.0312	22.0568	0.6278	0.6095	0.6274	0.0	0.6278	0.6276
2.6569	4	13092	2.5363	0.0625	34.8776	0.6421	0.6170	0.6426	0.0	0.6421	0.6419
2.321	5	16365	2.3178	0.125	61.7988	0.7090	0.7022	0.7093	0.0	0.7096	0.7088
2.2317	6	19638	2.1654	0.25	113.5221	0.7248	0.7189	0.7243	0.0	0.7248	0.7244
1.9634	7	22911	1.9102	0.5	219.7139	0.7739	0.7739	0.7739	0.0	0.7743	0.7736
1.8663	8.0	26184	2.0647	1.0	428.5444	0.7607	0.7567	0.7605	0.0	0.7610	0.7607
1.5382	9.0	29457	1.9436	1.0	426.5917	0.7785	0.7785	0.7787	0.0	0.7790	0.7781
1.028	10.0	32730	2.1566	1.0	432.3576	0.7776	0.7776	0.7776	0.0	0.7779	0.7774
0.8173	11.0	36003	2.9518	1.0	432.6197	0.7688	0.7680	0.7686	0.0	0.7691	0.7686

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1