mt5-base-joint-full-ft

This model is a fine-tuned version of google/mt5-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.9072
• Exact Match: 0.4052
• Sentiment Accuracy: 0.4988
• Sentiment Macro F1: 0.4877
• Stance Accuracy: 0.5948
• Stance Macro F1: 0.5799
• Avg Macro F1: 0.5338

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: 16
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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: 300
• num_epochs: 10
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Exact Match	Sentiment Accuracy	Sentiment Macro F1	Stance Accuracy	Stance Macro F1	Avg Macro F1
15.6789	1.0	189	10.4932	0.0	0.0	0.0	0.0	0.0	0.0
3.3977	2.0	378	2.1857	0.1708	0.3204	0.2288	0.3603	0.2775	0.2531
2.0704	3.0	567	1.9701	0.3142	0.3853	0.3281	0.5387	0.5180	0.4230
2.0195	4.0	756	1.9375	0.3392	0.4289	0.4106	0.5611	0.5455	0.4780
2.0105	5.0	945	1.9215	0.3716	0.4638	0.4451	0.5873	0.5758	0.5105
2.0005	6.0	1134	1.9111	0.3865	0.4825	0.4749	0.5973	0.5848	0.5299
1.9920	7.0	1323	1.9094	0.3978	0.4850	0.4705	0.5910	0.5730	0.5217
1.9866	8.0	1512	1.9076	0.4090	0.4975	0.4822	0.6010	0.5839	0.5331
1.9920	9.0	1701	1.9069	0.4002	0.4950	0.4847	0.5935	0.5795	0.5321
1.9890	10.0	1890	1.9072	0.4052	0.4988	0.4877	0.5948	0.5799	0.5338

Framework versions

• Transformers 5.16.1
• Pytorch 2.11.0+cu128
• Datasets 4.8.5
• Tokenizers 0.23.1