longt5_xl_summ_screen_memsum_bp_30

This model is a fine-tuned version of longt5_xl_summ_screen_memsum_bp_20/checkpoint-140 on the learn3r/summ_screen_fd_memsum_bp dataset. It achieves the following results on the evaluation set:

• Loss: 2.6817
• Rouge1: 47.1842
• Rouge2: 18.22
• Rougel: 28.4626
• Rougelsum: 45.5778
• Gen Len: 308.9083

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
• gradient_accumulation_steps: 32
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
0.0707	0.97	14	2.7097	41.4751	15.5831	25.1976	39.9229	453.5296
0.0608	1.95	28	2.7271	45.691	17.905	27.9519	43.8787	387.4172
0.0851	2.99	43	3.0001	47.1647	17.8993	28.7561	45.661	261.5680
0.0697	3.97	57	2.9297	46.6892	17.8922	28.0724	44.8821	365.3047
0.0296	4.94	71	2.9017	44.2702	17.7874	26.7598	42.6857	440.6391
0.0312	5.98	86	3.0489	47.7884	18.1788	28.6688	46.0744	306.6716
0.0383	6.96	100	2.6817	47.1842	18.22	28.4626	45.5778	308.9083
0.0367	8.0	115	3.0245	45.5573	17.2161	28.0573	43.7772	227.8550
0.04	8.97	129	3.2873	44.0164	17.1682	26.4769	42.3752	429.8757
0.028	9.74	140	2.9815	46.6542	17.8515	28.146	45.0274	337.4822

Framework versions

• Transformers 4.34.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3