learning_rate: 0.0003
train_batch_size: 128
eval_batch_size: 128
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 256
optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: inverse_sqrt
lr_scheduler_warmup_steps: 8000
num_epochs: 10
label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Chrf	Gen Len
15.3114	0.1148	2000	7.5526	0.3798	10.781	127.0
11.6538	0.2296	4000	5.4532	4.4773	25.6875	58.9224
9.1924	0.3445	6000	4.3517	9.1151	36.3059	36.6315
8.3172	0.4593	8000	3.9632	12.0107	39.6421	40.9618
7.7347	0.5741	10000	3.6736	14.5443	42.359	37.2294
7.3219	0.6889	12000	3.5034	16.4317	44.0952	36.8227
7.0472	0.8037	14000	3.3783	17.9723	46.0555	39.5261
6.8362	0.9186	16000	3.2669	19.1662	47.165	36.9328
6.6218	1.0334	18000	3.1924	20.8058	48.1287	38.9710
6.4850	1.1482	20000	3.1457	21.0326	48.6305	37.5979
6.3948	1.2630	22000	3.0906	21.4722	48.9497	36.1912
6.3099	1.3778	24000	3.0665	22.9794	50.1652	37.3778
6.2460	1.4926	26000	3.0223	22.8858	49.9641	37.6744
6.1850	1.6075	28000	2.9911	23.1469	50.6156	36.7462
6.1272	1.7223	30000	2.9886	23.7444	51.0271	37.1958
6.0775	1.8371	32000	2.9556	23.8764	50.9297	36.5979
6.0334	1.9519	34000	2.9273	25.4098	51.9017	36.6744
5.8709	2.0667	36000	2.9126	25.2662	52.1812	37.9328
5.8711	2.1815	38000	2.8997	25.1416	51.9853	36.6744
5.8527	2.2964	40000	2.8828	25.6802	52.5512	37.0811
5.8321	2.4112	42000	2.8662	25.3716	52.2047	37.1912
5.8096	2.5260	44000	2.8545	25.6273	52.4418	37.8946
5.7882	2.6408	46000	2.8543	26.1349	52.9929	37.4114
5.7684	2.7556	48000	2.8205	26.4363	53.3924	36.9664
5.7501	2.8705	50000	2.8160	27.1292	53.4956	37.8227
5.7336	2.9853	52000	2.8162	26.544	53.2085	38.4542
5.6070	3.1001	54000	2.8178	26.8796	53.4375	39.0046
5.6057	3.2149	56000	2.8008	27.0065	53.5326	36.8563
5.6023	3.3297	58000	2.7851	27.1542	53.5006	38.3395
5.5990	3.4445	60000	2.7909	27.0665	53.5857	38.5597
5.5951	3.5593	62000	2.7780	27.0759	53.5168	38.3778
5.5826	3.6742	64000	2.7747	27.504	54.0484	37.8227
5.5766	3.7890	66000	2.7720	27.5317	54.1437	38.0093
5.5671	3.9038	68000	2.7590	27.7362	54.0257	38.3059
5.5157	4.0186	70000	2.7634	27.9822	54.4644	36.8227
5.4430	4.1334	72000	2.7551	27.7185	54.1158	38.4878
5.4518	4.2482	74000	2.7523	28.4761	54.5614	37.8563
5.4538	4.3631	76000	2.7450	28.2706	54.7321	37.5214
5.4535	4.4779	78000	2.7491	28.4611	54.4191	37.5597
5.4538	4.5927	80000	2.7365	27.9624	54.4126	37.4496
5.4548	4.7075	82000	2.7188	28.7067	54.5294	37.2294
5.4476	4.8223	84000	2.7257	28.6484	54.6074	36.3013
5.4376	4.9372	86000	2.7236	28.6869	54.7994	37.5261
5.3236	5.0520	88000	2.7222	28.6419	54.976	38.6744
5.3359	5.1668	90000	2.7177	28.5266	54.8334	36.9710
5.3430	5.2816	92000	2.7144	28.2914	54.8561	37.4878
5.3456	5.3964	94000	2.7165	28.7312	55.0545	36.8227
5.3448	5.5112	96000	2.7037	28.4145	54.7407	37.7845
5.3552	5.6261	98000	2.7103	29.4486	55.278	37.7462
5.3526	5.7409	100000	2.7127	28.801	54.9646	38.4878
5.3517	5.8557	102000	2.7003	29.0762	55.2533	37.7080
5.3507	5.9705	104000	2.6947	29.1968	55.2374	38.0093
5.2448	6.0853	106000	2.7004	29.2967	55.1747	36.7845
5.2526	6.2001	108000	2.6948	29.1562	55.2315	38.4878
5.2601	6.3150	110000	2.6999	29.4735	55.5081	37.7462
5.2720	6.4298	112000	2.6928	28.9911	55.0498	37.7080
5.2777	6.5446	114000	2.6920	29.6562	55.4142	38.0811
5.2761	6.6594	116000	2.6817	29.0561	55.2792	37.0429
5.2729	6.7742	118000	2.6852	29.1892	55.1699	37.0811
5.2832	6.8891	120000	2.6769	28.997	55.1977	38.0093
5.2701	7.0038	122000	2.6877	29.0789	55.1696	37.7126
5.1779	7.1187	124000	2.6873	29.2204	55.3161	37.6362
5.1866	7.2335	126000	2.6780	29.5141	55.2472	38.0429
5.2074	7.3483	128000	2.6748	29.6794	55.5231	38.2630
5.2044	7.4631	130000	2.6751	29.3888	55.4729	38.4160
5.2087	7.5779	132000	2.6770	29.3841	55.3863	36.6744
5.2190	7.6928	134000	2.6803	29.035	55.3936	38.5261
5.2207	7.8076	136000	2.6735	29.2547	55.2594	37.7126
5.2225	7.9224	138000	2.6665	29.324	55.6016	37.2677
5.1466	8.0372	140000	2.6641	29.5036	55.4835	38.2677
5.1225	8.1520	142000	2.6676	29.8431	55.7386	37.5643
5.1401	8.2668	144000	2.6719	29.7275	55.4219	39.0429
5.1472	8.3817	146000	2.6618	29.8157	55.7436	36.4496
5.1575	8.4965	148000	2.6755	30.0132	55.9124	37.9328
5.1576	8.6113	150000	2.6609	30.0136	55.8688	37.0811
5.1658	8.7261	152000	2.6572	30.0187	55.6733	37.3395
5.1640	8.8409	154000	2.6592	29.9868	55.8766	38.7080
5.1676	8.9558	156000	2.6601	29.4069	55.5662	37.6744
5.0678	9.0706	158000	2.6620	29.616	55.7595	37.7462
5.0780	9.1854	160000	2.6649	29.8854	56.0274	37.7126
5.0895	9.3002	162000	2.6657	29.8918	56.0764	37.2294
5.1025	9.4150	164000	2.6630	29.2364	55.6724	38.3778
5.1070	9.5298	166000	2.6585	30.0675	55.8914	37.5643
5.1173	9.6447	168000	2.6487	30.433	56.1326	38.7462
5.1246	9.7595	170000	2.6550	30.0248	55.958	37.8227
5.1281	9.8743	172000	2.6506	29.818	55.7533	37.7462
5.1283	9.9891	174000	2.6502	30.06	55.9464	37.5979
5.1283	10.0	174190	2.6380	30.189	56.0749	36.8563

Framework versions

Transformers 5.6.2
Pytorch 2.5.1+cu121
Datasets 4.8.5
Tokenizers 0.22.2

alexantonov/chv-ru-marian-bt-3m-768d

model

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

alexantonov/
chv-ru-marian-bt-3m-768d