de83cf709744250dc83b322e48468e47

This model is a fine-tuned version of meta-llama/Llama-3.2-1B-Instruct on the dair-ai/emotion [split] dataset. It achieves the following results on the evaluation set:

• Loss: 1.2755
• Data Size: 1.0
• Epoch Runtime: 69.6963
• Accuracy: 0.9219
• F1 Macro: 0.8763

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
No log	0	0	14.8492	0	3.5393	0.1346	0.0843
No log	1	500	7.2937	0.0078	3.8074	0.3569	0.1065
No log	2	1000	7.0190	0.0156	5.0424	0.3488	0.0862
No log	3	1500	5.0949	0.0312	7.9655	0.5388	0.2828
No log	4	2000	2.5497	0.0625	10.0952	0.8281	0.7546
0.2318	5	2500	2.1216	0.125	16.1331	0.8624	0.8040
1.3367	6	3000	1.1815	0.25	23.9742	0.9158	0.8754
0.1455	7	3500	0.7878	0.5	41.0433	0.9244	0.8833
0.8008	8.0	4000	0.7646	1.0	74.4577	0.9209	0.8859
0.622	9.0	4500	1.0541	1.0	74.6973	0.9148	0.8725
0.7999	10.0	5000	0.6135	1.0	70.1165	0.9340	0.9020
0.5168	11.0	5500	0.8473	1.0	71.1821	0.9269	0.8897
0.533	12.0	6000	0.7022	1.0	72.4145	0.9304	0.8765
0.521	13.0	6500	0.8557	1.0	69.8117	0.9224	0.8807
0.4146	14.0	7000	1.2755	1.0	69.6963	0.9219	0.8763

Framework versions

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