mt5-small-synthetic-data-plus-translated

This model is a fine-tuned version of google/mt5-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5891
• Rouge1: 0.6390
• Rouge2: 0.5109
• Rougel: 0.6157
• Rougelsum: 0.6175

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: 5.6e-05
• train_batch_size: 12
• eval_batch_size: 12
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum
14.4747	1.0	100	4.4435	0.0225	0.0054	0.0205	0.0215
5.9023	2.0	200	1.9711	0.1865	0.0791	0.1562	0.1567
3.0374	3.0	300	1.3288	0.3668	0.2195	0.3565	0.3567
2.1905	4.0	400	1.1478	0.4430	0.2741	0.4186	0.4205
1.8996	5.0	500	1.0408	0.4754	0.3275	0.4564	0.4574
1.6959	6.0	600	0.9541	0.5463	0.3972	0.5258	0.5273
1.5593	7.0	700	0.8942	0.5594	0.4138	0.5406	0.5426
1.4334	8.0	800	0.8482	0.6064	0.4683	0.5855	0.5866
1.3929	9.0	900	0.8106	0.6130	0.4714	0.5895	0.5911
1.2918	10.0	1000	0.7851	0.6156	0.4770	0.5929	0.5935
1.2362	11.0	1100	0.7576	0.6270	0.4894	0.6054	0.6060
1.1781	12.0	1200	0.7402	0.6257	0.4867	0.6031	0.6042
1.1476	13.0	1300	0.7212	0.6221	0.4894	0.6018	0.6029
1.1052	14.0	1400	0.7064	0.6214	0.4873	0.5983	0.5995
1.0667	15.0	1500	0.6938	0.6300	0.4972	0.6073	0.6079
1.0421	16.0	1600	0.6855	0.6265	0.4952	0.6026	0.6036
1.0169	17.0	1700	0.6748	0.6244	0.4911	0.6021	0.6029
1.0036	18.0	1800	0.6599	0.6342	0.5087	0.6130	0.6142
0.9828	19.0	1900	0.6510	0.6349	0.5090	0.6136	0.6147
0.9589	20.0	2000	0.6471	0.6370	0.5074	0.6124	0.6135
0.9267	21.0	2100	0.6400	0.6345	0.5081	0.6117	0.6127
0.9361	22.0	2200	0.6318	0.6336	0.5066	0.6126	0.6140
0.8992	23.0	2300	0.6291	0.6346	0.5066	0.6122	0.6125
0.9029	24.0	2400	0.6224	0.6367	0.5103	0.6152	0.6166
0.8815	25.0	2500	0.6159	0.6374	0.5078	0.6141	0.6157
0.8914	26.0	2600	0.6133	0.6356	0.5109	0.6120	0.6138
0.8548	27.0	2700	0.6091	0.6371	0.5089	0.6125	0.6145
0.8683	28.0	2800	0.6047	0.6387	0.5131	0.6149	0.6169
0.8483	29.0	2900	0.6020	0.6368	0.5096	0.6121	0.6133
0.8409	30.0	3000	0.5996	0.6405	0.5118	0.6139	0.6159
0.8407	31.0	3100	0.5997	0.6398	0.5123	0.6159	0.6177
0.8338	32.0	3200	0.5970	0.6385	0.5096	0.6144	0.6164
0.801	33.0	3300	0.5947	0.6361	0.5078	0.6122	0.6141
0.833	34.0	3400	0.5941	0.6386	0.5111	0.6154	0.6172
0.7751	35.0	3500	0.5921	0.6368	0.5065	0.6129	0.6148
0.8281	36.0	3600	0.5906	0.6409	0.5125	0.6183	0.6199
0.7803	37.0	3700	0.5898	0.6377	0.5097	0.6143	0.6162
0.8139	38.0	3800	0.5896	0.6398	0.5116	0.6166	0.6185
0.7922	39.0	3900	0.5894	0.6388	0.5109	0.6156	0.6174
0.8269	40.0	4000	0.5891	0.6390	0.5109	0.6157	0.6175

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0