mt5-base-honda

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

• Loss: 0.5629
• Rouge1: 43.6523
• Rouge2: 32.203
• Rougel: 43.3772
• Rougelsum: 43.3022
• Gen Len: 17.73

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	42	4.6175	8.5469	3.7013	8.3757	8.2968	24.2255
10.4753	1.99	84	2.0552	10.5551	4.9989	10.2056	10.1735	15.6706
3.9727	2.99	126	1.3455	17.4897	9.2722	17.0361	17.0862	22.9496
2.111	3.99	168	1.1659	25.8241	16.5472	24.746	24.758	20.2997
1.6646	4.99	210	1.0557	26.4848	16.6274	25.1504	25.1065	16.5816
1.4307	5.98	252	0.9476	27.9191	19.2121	26.8901	26.8025	17.4866
1.4307	6.98	294	0.8371	31.0402	21.0973	30.2516	30.1885	19.4837
1.2842	8.0	337	0.7391	30.1165	20.5251	29.578	29.4895	15.362
1.1333	9.0	379	0.7287	34.5585	25.243	34.0959	33.8699	14.1573
1.1795	9.99	421	0.8753	26.8989	18.8627	26.5091	26.3887	36.1039
1.3298	10.99	463	0.7194	32.1116	24.2333	31.8284	31.7729	28.5015
1.0536	11.99	505	0.6241	35.7743	27.7948	35.5473	35.4622	26.7211
1.0536	12.99	547	0.6308	37.2689	28.103	36.8811	36.8093	21.8665
0.888	13.98	589	0.6370	38.8088	29.6802	38.4384	38.2694	20.3501
0.8153	14.98	631	0.6071	37.9373	29.6887	37.652	37.4779	22.8902
0.7717	16.0	674	0.5852	40.3825	29.9582	40.2962	40.1624	18.7389
0.734	17.0	716	0.5800	40.6092	30.1735	40.4011	40.3258	18.3442
0.6963	17.99	758	0.5797	39.7132	29.0489	39.5127	39.3232	21.0682
0.6574	18.99	800	0.5892	39.4966	29.6245	39.2659	39.1309	20.1068
0.6574	19.99	842	0.5715	40.7632	30.7816	40.3728	40.2779	18.0267
0.616	20.99	884	0.5648	41.988	31.7066	41.6728	41.6091	18.4955
0.5983	21.98	926	0.5699	42.1128	31.661	41.9032	41.7323	16.9466
0.5726	22.98	968	0.5636	41.4489	30.5531	41.1694	41.1125	19.6053
0.5577	24.0	1011	0.5603	43.0244	31.7556	42.7213	42.6249	16.9733
0.5405	25.0	1053	0.5715	41.9882	31.1594	41.7023	41.5209	18.1068
0.5405	25.99	1095	0.5587	42.7531	31.9549	42.4015	42.3466	17.5786
0.5355	26.99	1137	0.5702	42.0918	31.0895	41.6868	41.6741	17.27
0.5041	27.99	1179	0.5520	43.1863	32.0579	42.8749	42.7887	17.2433
0.5005	28.99	1221	0.5683	42.2837	31.0531	42.0168	41.9643	17.5668
0.5013	29.98	1263	0.5626	42.8554	31.6127	42.6408	42.5058	17.9318
0.4599	30.98	1305	0.5637	42.8309	31.2431	42.6061	42.4722	18.6617
0.4599	32.0	1348	0.5620	43.4879	32.0117	43.1924	43.1162	17.3709
0.4783	33.0	1390	0.5616	42.9605	31.3625	42.6897	42.5864	18.2522
0.4742	33.99	1432	0.5548	43.2898	31.6867	43.0984	42.9612	17.5015
0.4598	34.99	1474	0.5596	43.9791	32.2278	43.6851	43.5645	17.5757
0.4381	35.99	1516	0.5638	43.7052	32.2458	43.392	43.31	17.6261
0.4496	36.99	1558	0.5567	43.9806	32.2596	43.654	43.6041	17.8902
0.4464	37.98	1600	0.5615	43.7515	32.4353	43.5184	43.4019	17.4629
0.4464	38.98	1642	0.5625	43.7698	32.3006	43.4733	43.3957	17.5935
0.4431	39.88	1680	0.5629	43.6523	32.203	43.3772	43.3022	17.73

Framework versions

• Transformers 4.39.3
• Pytorch 2.2.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1