t5-qg-checkpoints

This model is a fine-tuned version of t5-base on the nl-quad dataset. It achieves the following results on the evaluation set:

• Loss: 2.0288
• Rougel: 42.2
• Bleu: 18.85
• Meteor: 41.13
• Bert Precision: 94.19
• Bert Recall: 93.51
• Bert F1: 93.84
• Qsts Mean: 63.5100
• Gen Len: 13.48

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: 48
• eval_batch_size: 48
• seed: 42
• 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: linear
• lr_scheduler_warmup_steps: 0.1
• num_epochs: 10
• label_smoothing_factor: 0.05

Training results

Training Loss	Epoch	Step	Validation Loss	Rougel	Bleu	Meteor	Bert Precision	Bert Recall	Bert F1	Qsts Mean	Gen Len
2.3343	1.0	249	2.1434	38.63	14.89	37.32	93.83	93.09	93.45	61.4900	12.94
2.1126	2.0	498	2.0469	40.24	16.75	39.0	94.01	93.27	93.63	63.2800	13.43
1.9399	3.0	747	2.0290	40.4	16.41	39.06	94.0	93.31	93.64	62.8100	13.7
1.8319	4.0	996	2.0196	41.67	17.83	40.27	94.13	93.46	93.78	63.4000	13.8
1.7543	5.0	1245	2.0204	41.41	18.1	40.28	94.07	93.5	93.77	63.9000	13.74
1.6840	6.0	1494	2.0288	42.2	18.85	41.13	94.19	93.51	93.84	63.5100	13.48

Framework versions

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