metadata

license: apache-2.0
base_model: google/mt5-small
tags:
  - generated_from_trainer
metrics:
  - rouge
model-index:
  - name: mt5-summarize-ja
    results: []

mt5-summarize-ja

This model is a fine-tuned version of google/mt5-small on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 2.0695
Rouge1: 0.3667
Rouge2: 0.1678
Rougel: 0.2998
Rougelsum: 0.3123

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: 0.0005
train_batch_size: 2
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 16
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 90
num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
3.3241	0.7	100	2.4795	0.2943	0.1245	0.2472	0.2471
2.7583	1.4	200	2.2710	0.3054	0.1152	0.2539	0.2576
2.5469	2.1	300	2.2936	0.3446	0.1493	0.2808	0.2887
2.5335	2.8	400	2.1913	0.3228	0.1270	0.2665	0.2725
2.4383	3.5	500	2.1507	0.3630	0.1671	0.3082	0.3144
2.3671	4.2	600	2.1338	0.3388	0.1493	0.2814	0.2880
2.349	4.9	700	2.1089	0.3621	0.1576	0.2980	0.3079
2.264	5.6	800	2.1353	0.3740	0.1784	0.3083	0.3157
2.1577	6.3	900	2.1101	0.3711	0.1716	0.3107	0.3166
2.1315	7.0	1000	2.0905	0.3862	0.1826	0.3198	0.3269
2.1418	7.7	1100	2.0893	0.3433	0.1621	0.2895	0.2963
2.0744	8.4	1200	2.0881	0.3778	0.1834	0.3130	0.3242
2.0944	9.1	1300	2.0709	0.3676	0.1688	0.3024	0.3140
2.1015	9.8	1400	2.0695	0.3667	0.1678	0.2998	0.3123

Framework versions

Transformers 4.35.2
Pytorch 2.1.0+cu121
Datasets 2.17.0
Tokenizers 0.15.2