story2dialogue-SODA-BART-LoRA

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

• Loss: 1.7296
• Rouge1: 0.2294
• Rouge2: 0.0795
• Rougel: 0.2019
• Rougelsum: 0.2019
• Bleu: 0.0447
• Bleu1: 0.3122
• Bleu2: 0.0914
• Bleu3: 0.0447
• Bleu4: 0.0252
• Meteor: 0.2076
• Avg Distinct 1: 0.9644
• Avg Distinct 2: 0.9983
• Avg Distinct 3: 0.9942
• Avg Jaccard: 0.0915
• Gen Length: 10.6426

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH 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: 500
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Bleu	Bleu1	Bleu2	Bleu3	Bleu4	Meteor	Avg Distinct 1	Avg Distinct 2	Avg Distinct 3	Avg Jaccard	Gen Length
2.1855	0.4008	1000	1.8555	0.1866	0.0530	0.1629	0.1629	0.0292	0.2853	0.0658	0.0292	0.0146	0.1717	0.9666	0.9987	0.9951	0.0677	9.9008
2.1	0.8016	2000	1.8014	0.2004	0.0604	0.1753	0.1752	0.0331	0.2943	0.0738	0.0339	0.0172	0.1842	0.9668	0.9983	0.9949	0.0741	9.9986
2.0609	1.2024	3000	1.7680	0.2105	0.0673	0.1846	0.1846	0.0377	0.3033	0.0819	0.0378	0.0197	0.1937	0.9623	0.9982	0.9928	0.0808	10.2733
2.0165	1.6032	4000	1.7527	0.2199	0.0730	0.1930	0.1930	0.0422	0.3045	0.0862	0.0419	0.0231	0.2004	0.9628	0.9981	0.9944	0.0866	10.6475
2.0087	2.0040	5000	1.7410	0.2250	0.0759	0.1977	0.1977	0.0437	0.3101	0.0900	0.0439	0.0242	0.2035	0.9642	0.9983	0.9957	0.0898	10.6315
2.0004	2.4048	6000	1.7333	0.2286	0.0784	0.2009	0.2010	0.0453	0.3092	0.0911	0.0447	0.0254	0.2062	0.9623	0.9978	0.9938	0.0911	10.7750
1.9937	2.8056	7000	1.7296	0.2294	0.0795	0.2019	0.2019	0.0447	0.3122	0.0914	0.0447	0.0252	0.2076	0.9644	0.9983	0.9942	0.0915	10.6426

Framework versions

• PEFT 0.15.2
• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.2