smartmind-cyberone-20250402

This model is a fine-tuned version of PowerInfer/SmallThinker-3B-Preview on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0308

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: 2e-05
• train_batch_size: 64
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 64
• total_train_batch_size: 4096
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine_with_restarts
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.3817	0.3527	30	0.2134
0.1292	0.7054	60	0.0959
0.0931	1.0470	90	0.3264
0.125	1.3997	120	0.0485
0.057	1.7524	150	0.0569
0.0503	2.0940	180	0.0444
0.0444	2.4467	210	0.0426
0.0405	2.7994	240	0.0346
0.0472	3.1411	270	0.0614
0.045	3.4938	300	0.0406
0.0405	3.8464	330	0.0328
0.0345	4.1881	360	0.0300
0.0333	4.5408	390	0.0363
0.0325	4.8935	420	0.0308

Framework versions

• Transformers 4.50.3
• Pytorch 2.5.1+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1