gpt-mqa-RoPE

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

• Loss: 6.0885

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: 128
• eval_batch_size: 128
• seed: 20
• gradient_accumulation_steps: 4
• total_train_batch_size: 512
• 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: 106
• training_steps: 1060
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
9.5007	0.0590	106	9.2225
7.6938	0.1179	212	7.6025
7.0385	0.1769	318	6.9495
6.6472	0.2359	424	6.5905
6.4123	0.2948	530	6.3711
6.2808	0.3538	636	6.2347
6.1927	0.4127	742	6.1516
6.1486	0.4717	848	6.1083
6.1331	0.5307	954	6.0914
6.1282	0.5896	1060	6.0885

Framework versions

• Transformers 5.5.4
• Pytorch 2.11.0+cu130
• Datasets 4.8.4
• Tokenizers 0.22.2