authorship_model

This model is a fine-tuned version of Qwen/Qwen2.5-1.5B-Instruct on the authorship_train dataset. It achieves the following results on the evaluation set:

• Loss: 0.5754

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: 0.0001
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 16
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 20
• num_epochs: 1.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.3515	0.0486	200	0.7383
1.2685	0.0972	400	0.7080
1.2484	0.1458	600	0.6671
1.2335	0.1945	800	0.6473
1.1783	0.2431	1000	0.6368
1.1715	0.2917	1200	0.6287
1.2187	0.3403	1400	0.6172
1.1708	0.3889	1600	0.6101
1.1623	0.4375	1800	0.6024
1.1197	0.4861	2000	0.6001
1.1420	0.5348	2200	0.5934
1.1861	0.5834	2400	0.5901
1.1843	0.6320	2600	0.5874
1.1776	0.6806	2800	0.5833
1.1156	0.7292	3000	0.5814
1.1301	0.7778	3200	0.5785
1.1356	0.8264	3400	0.5773
1.1167	0.8751	3600	0.5761
1.1620	0.9237	3800	0.5755
1.1024	0.9723	4000	0.5754

Framework versions

• PEFT 0.18.1
• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2