psxjp5/mt5-small_old

mt5-small_epochs_new_new

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: 1.0848
• Rouge1: 41.527
• Rouge2: 33.324
• Rougel: 38.4866
• Rougelsum: 38.4856
• Bleu: 29.906
• Gen Len: 17.1296
• Meteor: 0.377
• No ans accuracy: 47

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 9
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Bleu	Gen Len	Meteor	No ans accuracy
9.0096	1.0	316	2.5283	23.4252	15.294	21.7032	21.7597	9.5703	12.1099	0.2117	0
3.2564	2.0	632	1.8337	33.2328	25.7922	31.2553	31.2709	15.51	13.8804	0.3108	0
2.5244	3.0	948	1.5796	34.1863	26.9908	32.3162	32.3197	17.5684	14.1904	0.3262	0
2.1686	3.99	1264	1.4179	34.565	27.4829	32.6012	32.6306	18.1896	14.2814	0.329	0
1.9465	4.99	1580	1.3050	41.2984	32.8587	38.3901	38.3985	28.3953	17.1212	0.3724	24
1.8009	5.99	1896	1.2428	41.5784	33.0684	38.6495	38.6555	28.9287	17.2045	0.3755	27
1.6954	6.99	2212	1.1992	40.4868	32.2937	37.6021	37.5986	28.2477	16.8056	0.3662	54
1.6322	7.99	2528	1.1769	37.6427	30.0271	34.8637	34.8951	26.433	15.5656	0.34	124
1.5845	8.99	2844	1.1574	40.3396	32.2547	37.3672	37.4137	28.6687	16.6457	0.3638	66
1.5425	9.98	3160	1.1500	39.1906	31.3426	36.3113	36.3654	27.8135	16.1679	0.3542	95
1.5137	10.98	3476	1.1367	41.4173	33.1848	38.4473	38.4306	29.6548	17.0306	0.3755	51
1.4826	11.98	3792	1.1161	41.4856	33.1913	38.4806	38.4896	29.5512	17.1031	0.3762	44
1.4514	12.98	4108	1.1182	41.8374	33.5091	38.7582	38.7679	29.8577	17.2987	0.3797	37
1.4444	13.98	4424	1.1056	42.0345	33.6905	38.9576	38.9795	30.1371	17.2669	0.3823	38
1.425	14.98	4740	1.0973	41.5086	33.2216	38.4098	38.4115	29.7019	17.1244	0.3767	50
1.407	15.97	5056	1.0890	41.7122	33.4259	38.605	38.6225	29.9984	17.1908	0.3794	44
1.4005	16.97	5372	1.0881	41.5731	33.2998	38.521	38.5259	29.9097	17.1027	0.3775	49
1.3865	17.97	5688	1.0860	40.9767	32.8412	37.9532	37.9637	29.4171	16.9404	0.372	55
1.3849	18.97	6004	1.0848	41.527	33.324	38.4866	38.4856	29.906	17.1296	0.377	47

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3