TURNA_spell_correction_general_turkish

This model is a fine-tuned version of boun-tabi-LMG/TURNA on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1739
• Rouge1: 0.6731
• Rouge2: 0.3509
• Rougel: 0.6734
• Rougelsum: 0.6734
• Bleu: 0.0
• Precisions: [0.724791169451074, 0.875, 0.9080188679245284, 0.0]
• Brevity Penalty: 0.9975
• Length Ratio: 0.9975
• Translation Length: 6704
• Reference Length: 6721
• Meteor: 0.4269
• Score: 27.9423
• Num Edits: 1878
• Ref Length: 6721.0

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: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Bleu	Precisions	Brevity Penalty	Length Ratio	Translation Length	Reference Length	Meteor	Score	Num Edits	Ref Length
No log	0.3334	1407	0.3554	0.5067	0.2640	0.5068	0.5068	0.0	[0.5543951692734113, 0.681206764027671, 0.620919881305638, 0.0]	1.0	1.0039	20204	20126	0.3130	45.8213	9222	20126.0
No log	0.6668	2814	0.2851	0.5589	0.2938	0.5590	0.5591	0.0	[0.612387263939409, 0.7749062931544684, 0.7614457831325301, 0.0]	0.9972	0.9972	20069	20126	0.3509	39.5906	7968	20126.0
0.5458	1.0002	4221	0.2465	0.5908	0.3104	0.5914	0.5912	0.0	[0.6474730323611666, 0.8186703821656051, 0.8253452477660439, 0.0]	0.9949	0.9949	20024	20126	0.3739	36.0330	7252	20126.0
0.5458	1.3336	5628	0.2302	0.6130	0.3223	0.6133	0.6135	0.0	[0.6655887766777773, 0.8296412468143501, 0.8471337579617835, 0.0]	0.9988	0.9988	20101	20126	0.3876	34.1002	6863	20126.0
0.5458	1.6671	7035	0.2101	0.6242	0.3262	0.6241	0.6246	0.0	[0.6759966072943172, 0.8334324806662701, 0.8369652945924132, 0.0]	0.9959	0.9959	20043	20126	0.3936	33.0965	6661	20126.0
0.1935	2.0005	8442	0.1931	0.6498	0.3415	0.6501	0.6502	0.0	[0.6969411063660741, 0.8487738419618529, 0.8557993730407524, 0.0]	1.0	1.0006	20138	20126	0.4097	30.8457	6208	20126.0
0.1935	2.3339	9849	0.1939	0.6500	0.3443	0.6502	0.6502	0.0	[0.7011294094233544, 0.8635026475779565, 0.8777602523659306, 0.0]	0.9987	0.9987	20099	20126	0.4120	30.3886	6116	20126.0
0.1935	2.6673	11256	0.1864	0.6615	0.3467	0.6615	0.6621	0.0	[0.7094416546512206, 0.8591824760414629, 0.8735271013354281, 0.0]	0.9994	0.9994	20113	20126	0.4178	29.5439	5946	20126.0
0.1135	3.0007	12663	0.1746	0.6729	0.3488	0.6734	0.6735	0.0	[0.7206584444002387, 0.8674628034455756, 0.8738244514106583, 0.0]	0.9991	0.9991	20108	20126	0.4253	28.3961	5715	20126.0
0.1135	3.3341	14070	0.1846	0.6750	0.3540	0.6753	0.6753	0.0	[0.7217659648598875, 0.8744843842074249, 0.8928571428571429, 0.0]	0.9983	0.9983	20091	20126	0.4260	28.2918	5694	20126.0
0.1135	3.6675	15477	0.1806	0.6827	0.3566	0.6830	0.6829	0.0	[0.7288599283724632, 0.8759796238244514, 0.8909090909090909, 0.0]	0.9989	0.9989	20104	20126	0.4308	27.5465	5544	20126.0
0.0672	4.0009	16884	0.1758	0.6872	0.3617	0.6873	0.6874	0.0	[0.7345353122820998, 0.8853879480110279, 0.9000793021411578, 0.0]	0.9976	0.9976	20078	20126	0.4341	26.9701	5428	20126.0

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.1+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1