t5_chv_ru_model

This model is a fine-tuned version of slone/nllb-210-v1 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.3846
• Bleu: 24.5491
• Gen Len: 23.3803

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Use 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: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
1.6697	0.0548	5000	1.5639	21.3268	23.3023
1.6356	0.1096	10000	1.5370	21.7019	23.4241
1.6152	0.1644	15000	1.5185	21.7427	23.4938
1.5952	0.2192	20000	1.5045	22.3625	23.459
1.5816	0.2740	25000	1.4906	22.6058	23.3269
1.5493	0.3288	30000	1.4819	22.8262	23.3529
1.5598	0.3836	35000	1.4716	22.8165	23.3639
1.5490	0.4383	40000	1.4641	22.8553	23.3051
1.5494	0.4931	45000	1.4561	22.8919	23.2585
1.5260	0.5479	50000	1.4514	22.9107	23.4959
1.5192	0.6027	55000	1.4453	22.936	23.3003
1.5176	0.6575	60000	1.4378	23.0299	23.4631
1.5303	0.7123	65000	1.4341	23.1948	23.2914
1.5193	0.7671	70000	1.4295	23.3968	23.3181
1.5107	0.8219	75000	1.4250	23.5299	23.2873
1.5027	0.8767	80000	1.4213	23.5169	23.4685
1.4910	0.9315	85000	1.4183	23.2081	23.2818
1.4922	0.9863	90000	1.4136	23.6995	23.4733
1.4505	1.0411	95000	1.4120	23.4069	23.3071
1.4617	1.0959	100000	1.4109	23.7334	23.4268
1.4379	1.1507	105000	1.4078	23.6355	23.381
1.4324	1.2055	110000	1.4049	24.0022	23.2613
1.4475	1.2602	115000	1.4006	23.887	23.487
1.4308	1.3150	120000	1.3983	23.9695	23.2975
1.4276	1.3698	125000	1.3983	23.7433	23.3721
1.4359	1.4246	130000	1.3958	23.9696	23.3372
1.4259	1.4794	135000	1.3933	23.8664	23.37
1.4372	1.5342	140000	1.3928	24.11	23.3741
1.4334	1.5890	145000	1.3901	24.0618	23.3167
1.4324	1.6438	150000	1.3891	24.2506	23.3871
1.4297	1.6986	155000	1.3882	24.1492	23.3926
1.4296	1.7534	160000	1.3872	24.2394	23.3523
1.4315	1.8082	165000	1.3861	24.3765	23.3639
1.4295	1.8630	170000	1.3854	24.4829	23.3687
1.4382	1.9178	175000	1.3851	24.4165	23.3694
1.4243	1.9726	180000	1.3846	24.4305	23.37
1.4188	2.0	182504	1.3846	24.5491	23.3803

Framework versions

• Transformers 5.6.2
• Pytorch 2.5.1+cu121
• Datasets 4.8.5
• Tokenizers 0.22.2