KGAQ-2

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

• Loss: 6.0725
• Rouge1: 46.2376
• Rouge2: 21.4997
• Rougel: 39.6036
• Rougelsum: 46.3269
• Gen Len: 4.2121
• F1: 0.3205
• Recall: 0.6757
• Precision: 0.2101

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: 0.001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine_with_restarts
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len	F1	Recall	Precision
No log	1.0	50	3.1330	38.9228	19.8713	34.8665	39.0223	3.6162	0.3373	0.5957	0.2353
No log	2.0	100	3.2460	42.6051	20.0714	38.2234	42.823	3.9697	0.3275	0.5385	0.2353
No log	3.0	150	3.4413	42.2575	19.3868	36.9508	42.1996	4.1313	0.3415	0.6222	0.2353
1.9251	4.0	200	3.6553	41.9902	19.8751	36.961	42.1914	3.9899	0.3522	0.7	0.2353
1.9251	5.0	250	3.9188	41.6177	19.8385	36.9836	41.7831	4.0404	0.3648	0.725	0.2437
1.9251	6.0	300	4.0309	40.2818	15.9608	35.0963	40.3224	4.3838	0.3522	0.7	0.2353
1.9251	7.0	350	4.4151	40.1585	14.4247	34.3216	40.2886	4.3131	0.1185	0.5	0.0672
0.6344	8.0	400	4.9239	42.9643	19.2829	36.6803	43.0145	4.4646	0.3097	0.6667	0.2017
0.6344	9.0	450	5.9057	45.7386	21.5407	39.3743	45.7904	4.5253	0.3205	0.6757	0.2101
0.6344	10.0	500	6.0725	46.2376	21.4997	39.6036	46.3269	4.2121	0.3205	0.6757	0.2101

Framework versions

• Transformers 4.43.3
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1