all_8657_bart-base

This model is a fine-tuned version of facebook/bart-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.2356
• Rouge1: 0.2722
• Rouge2: 0.1239
• Rougel: 0.2315
• Rougelsum: 0.2424
• Gen Len: 19.9567

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 16
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.759	0.89	500	1.2348	0.2615	0.1071	0.2187	0.2283	19.956
1.0891	1.78	1000	1.2122	0.2667	0.1145	0.224	0.2351	19.9713
0.9877	2.67	1500	1.2076	0.2701	0.118	0.2271	0.238	19.9413
0.9299	3.56	2000	1.2072	0.2682	0.1205	0.2267	0.2385	19.9667
0.8841	4.44	2500	1.2088	0.2711	0.1213	0.2294	0.2406	19.956
0.8425	5.33	3000	1.2154	0.2718	0.1245	0.2317	0.2426	19.9673
0.8123	6.22	3500	1.2276	0.2719	0.1242	0.2315	0.2422	19.958
0.7876	7.11	4000	1.2259	0.2726	0.1228	0.2311	0.242	19.9647
0.769	8.0	4500	1.2244	0.2733	0.126	0.2324	0.2436	19.9667
0.75	8.89	5000	1.2313	0.2723	0.1236	0.231	0.2422	19.964
0.7369	9.78	5500	1.2356	0.2722	0.1239	0.2315	0.2424	19.9567

Framework versions

• Transformers 4.38.2
• Pytorch 2.0.0+cu117
• Datasets 2.18.0
• Tokenizers 0.15.2