flan-t5-base-srbd

This model is a fine-tuned version of google/flan-t5-base on an srbd1_v2_annotated_segmented. It achieves the following results on the evaluation set:

• Loss: 0.2061
• Rouge1: 73.9926
• Rouge2: 65.6762
• Rougel: 73.0659
• Rougelsum: 73.9428
• Gen Len: 15.7754

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: 8
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
11.6626	0.86	200	0.5153	51.3405	41.8375	50.9028	51.2567	7.5616
0.4412	1.72	400	0.2881	73.5628	64.0434	72.4555	73.42	15.7495
0.3127	2.59	600	0.2365	73.3305	64.0389	72.3218	73.2283	15.6328
0.2683	3.45	800	0.2397	73.6254	64.704	72.6818	73.5721	15.8272
0.2462	4.31	1000	0.2197	73.8635	65.1757	72.8344	73.6793	15.7235
0.2206	5.17	1200	0.2142	74.0454	65.5104	73.1022	73.9321	15.8553
0.2073	6.03	1400	0.2087	73.8199	65.2791	72.8211	73.7232	15.743
0.1951	6.9	1600	0.2109	74.0189	65.1189	73.0456	73.9537	15.6782
0.1946	7.76	1800	0.2078	74.1418	65.7362	73.3347	74.0834	15.7775
0.1885	8.62	2000	0.2062	74.3159	65.8017	73.2801	74.0865	15.7905
0.1793	9.48	2200	0.2061	73.9926	65.6762	73.0659	73.9428	15.7754

Framework versions

• Transformers 4.32.1
• Pytorch 2.0.0
• Datasets 2.1.0
• Tokenizers 0.13.3