chessgpt-medium-l

This model is a fine-tuned version of gpt2-medium on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7634

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: 0.0003
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.2
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.4326	0.064	1000	1.3596
1.2424	0.128	2000	1.1829
1.1278	0.192	3000	1.0753
1.0296	0.256	4000	0.9877
0.9605	0.32	5000	0.9224
0.9193	0.384	6000	0.8874
0.8911	0.448	7000	0.8600
0.8707	0.512	8000	0.8405
0.8521	0.576	9000	0.8221
0.8391	0.64	10000	0.8089
0.8242	0.704	11000	0.7972
0.8146	0.768	12000	0.7858
0.8047	0.832	13000	0.7769
0.7974	0.896	14000	0.7701
0.7916	0.96	15000	0.7651

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1