mt5-lithuanian-simplifier-full

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

• Loss: 0.0771
• Rouge1: 0.7828
• Rouge2: 0.6494
• Rougel: 0.7787
• Gen Len: 48.0191

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 8

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Gen Len
24.351	0.08	200	18.6244	0.0226	0.0018	0.0207	512.0
3.0331	0.16	400	0.6830	0.0549	0.0018	0.0497	49.0191
0.2076	0.24	600	0.1642	0.6417	0.4986	0.6328	48.0191
0.2019	0.32	800	0.1303	0.6713	0.5243	0.6633	48.0191
0.1573	0.4	1000	0.1242	0.7007	0.5589	0.6937	48.0191
0.1687	0.48	1200	0.1158	0.712	0.569	0.7055	48.0191
0.1315	0.56	1400	0.1225	0.6923	0.5361	0.6851	48.0191
0.1376	0.64	1600	0.1108	0.7171	0.5695	0.7105	48.0191
0.158	0.72	1800	0.1074	0.7229	0.574	0.7169	48.0191
0.1221	0.8	2000	0.1064	0.7227	0.5761	0.7166	48.0191
0.1371	0.88	2200	0.1049	0.7282	0.5827	0.7223	48.0191
0.1376	0.96	2400	0.1043	0.73	0.5861	0.7239	48.0191
0.1116	1.04	2600	0.1021	0.733	0.5888	0.727	48.0191
0.132	1.12	2800	0.1012	0.7338	0.5899	0.7277	48.0191
0.131	1.2	3000	0.0997	0.7365	0.5936	0.7307	48.0191
0.1001	1.28	3200	0.0950	0.7408	0.5977	0.7355	48.0191
0.1398	1.36	3400	0.0964	0.7418	0.599	0.7364	48.0191
0.1085	1.44	3600	0.0962	0.744	0.6015	0.7386	48.0191
0.097	1.52	3800	0.0967	0.743	0.6009	0.7377	48.0191
0.1178	1.6	4000	0.0955	0.7446	0.6035	0.7391	48.0191
0.1214	1.68	4200	0.0939	0.7452	0.6036	0.7403	48.0191
0.1539	1.76	4400	0.0909	0.7486	0.6068	0.7436	48.0191
0.1141	1.83	4600	0.0900	0.7518	0.6104	0.7467	48.0191
0.0795	1.91	4800	0.0891	0.7513	0.6097	0.7466	48.0191
0.0856	1.99	5000	0.0915	0.7513	0.6099	0.7463	48.0191
0.0954	2.07	5200	0.0898	0.753	0.6126	0.7482	48.0191
0.1271	2.15	5400	0.0901	0.7534	0.6125	0.7486	48.0191
0.0816	2.23	5600	0.0893	0.7553	0.6148	0.7506	48.0191
0.0922	2.31	5800	0.0881	0.7569	0.6163	0.7521	48.0191
0.1177	2.39	6000	0.0878	0.7575	0.6176	0.7532	48.0191
0.0916	2.47	6200	0.0874	0.7585	0.618	0.7541	48.0191
0.1349	2.55	6400	0.0861	0.76	0.62	0.7555	48.0191
0.1196	2.63	6600	0.0833	0.7617	0.6212	0.7572	48.0191
0.0841	2.71	6800	0.0848	0.7621	0.6219	0.7576	48.0191
0.0934	2.79	7000	0.0854	0.7622	0.6227	0.7577	48.0191
0.1246	2.87	7200	0.0835	0.7652	0.6256	0.7606	48.0191
0.0762	2.95	7400	0.0835	0.7649	0.6262	0.7606	48.0191
0.0924	3.03	7600	0.0828	0.7662	0.6276	0.7618	48.0191
0.0822	3.11	7800	0.0834	0.7664	0.6284	0.7621	48.0191
0.0856	3.19	8000	0.0836	0.7647	0.627	0.7603	48.0191
0.0798	3.27	8200	0.0829	0.7657	0.6284	0.7614	48.0191
0.0959	3.35	8400	0.0828	0.7671	0.6302	0.7629	48.0191
0.0871	3.43	8600	0.0820	0.7672	0.6297	0.763	48.0191
0.1068	3.51	8800	0.0827	0.7683	0.6307	0.7641	48.0191
0.072	3.59	9000	0.0820	0.7684	0.632	0.764	48.0191
0.0964	3.67	9200	0.0838	0.7692	0.6333	0.7645	48.0191
0.0946	3.75	9400	0.0809	0.7707	0.6348	0.7663	48.0191
0.0822	3.83	9600	0.0825	0.7708	0.6347	0.7666	48.0191
0.1019	3.91	9800	0.0788	0.7733	0.6373	0.7692	48.0191
0.08	3.99	10000	0.0797	0.7727	0.6369	0.7686	48.0191
0.0989	4.07	10200	0.0818	0.7724	0.6367	0.7681	48.0191
0.0693	4.15	10400	0.0804	0.7737	0.6378	0.7697	48.0191
0.0763	4.23	10600	0.0814	0.7741	0.6379	0.7699	48.0191
0.0956	4.31	10800	0.0815	0.7726	0.6369	0.7683	48.0191
0.0728	4.39	11000	0.0800	0.7738	0.6374	0.7696	48.0191
0.0652	4.47	11200	0.0795	0.7747	0.6388	0.7708	48.0191
0.0706	4.55	11400	0.0798	0.7742	0.6388	0.7703	48.0191
0.0979	4.63	11600	0.0788	0.7748	0.6387	0.7708	48.0191
0.0771	4.71	11800	0.0797	0.775	0.6402	0.771	48.0191
0.1067	4.79	12000	0.0779	0.7757	0.6404	0.7717	48.0191
0.0773	4.87	12200	0.0783	0.7759	0.6411	0.7721	48.0191
0.0866	4.95	12400	0.0780	0.7773	0.6437	0.7734	48.0191
0.0611	5.03	12600	0.0785	0.7761	0.6418	0.7723	48.0191
0.0685	5.11	12800	0.0781	0.777	0.6421	0.773	48.0191
0.0501	5.19	13000	0.0788	0.7764	0.6411	0.7721	48.0191
0.0626	5.27	13200	0.0792	0.7762	0.6416	0.7721	48.0191
0.0708	5.35	13400	0.0795	0.7761	0.6408	0.772	48.0191
0.055	5.42	13600	0.0779	0.7773	0.642	0.7733	48.0191
0.0749	5.5	13800	0.0789	0.7783	0.6431	0.7742	48.0191
0.0771	5.58	14000	0.0779	0.778	0.6437	0.774	48.0191
0.0906	5.66	14200	0.0779	0.7781	0.6431	0.7742	48.0191
0.0679	5.74	14400	0.0778	0.7783	0.6449	0.7745	48.0191
0.0605	5.82	14600	0.0786	0.7778	0.6439	0.7738	48.0191
0.0647	5.9	14800	0.0781	0.7785	0.6445	0.7743	48.0191
0.058	5.98	15000	0.0775	0.7792	0.6448	0.7749	48.0191
0.0574	6.06	15200	0.0788	0.7793	0.6451	0.7752	48.0191
0.0545	6.14	15400	0.0778	0.7802	0.6464	0.7759	48.0191
0.079	6.22	15600	0.0781	0.7801	0.6466	0.7759	48.0191
0.0474	6.3	15800	0.0782	0.7809	0.6477	0.7768	48.0191
0.0517	6.38	16000	0.0788	0.7809	0.6481	0.7769	48.0191
0.0613	6.46	16200	0.0782	0.7814	0.6481	0.7773	48.0191
0.0517	6.54	16400	0.0785	0.7807	0.6468	0.7767	48.0191
0.0549	6.62	16600	0.0778	0.7817	0.6485	0.7777	48.0191
0.0727	6.7	16800	0.0774	0.7824	0.6493	0.7785	48.0191
0.0768	6.78	17000	0.0784	0.7826	0.6495	0.7785	48.0191
0.0612	6.86	17200	0.0772	0.7818	0.6485	0.7779	48.0191
0.0735	6.94	17400	0.0778	0.7817	0.6484	0.7777	48.0191
0.0662	7.02	17600	0.0780	0.7819	0.6483	0.7778	48.0191
0.0769	7.1	17800	0.0777	0.7823	0.6488	0.7784	48.0191
0.0649	7.18	18000	0.0775	0.7818	0.6482	0.7778	48.0191
0.0749	7.26	18200	0.0774	0.7822	0.6486	0.7781	48.0191
0.0568	7.34	18400	0.0772	0.7825	0.6488	0.7784	48.0191
0.0751	7.42	18600	0.0774	0.7822	0.6486	0.7783	48.0191
0.0564	7.5	18800	0.0773	0.7823	0.6487	0.7782	48.0191
0.0593	7.58	19000	0.0767	0.7826	0.6492	0.7786	48.0191
0.0563	7.66	19200	0.0773	0.7826	0.6497	0.7786	48.0191
0.0686	7.74	19400	0.0771	0.7828	0.6494	0.7789	48.0191
0.0728	7.82	19600	0.0772	0.7823	0.6494	0.7784	48.0191
0.06	7.9	19800	0.0772	0.7826	0.6491	0.7786	48.0191
0.0557	7.98	20000	0.0771	0.7828	0.6494	0.7787	48.0191

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.1
• Datasets 2.16.1
• Tokenizers 0.15.0