char-text-reversal

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

• Loss: 1.0816
• Char Accuracy: 0.0065
• Sequence Accuracy: 0.0
• Edit Distance: 38.583

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: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 1000

Training results

Training Loss	Epoch	Step	Validation Loss	Char Accuracy	Sequence Accuracy	Edit Distance
4.1717	1.0	79	3.7332	0.0319	0.0	131.1735
3.4932	2.0	158	3.2892	0.0011	0.0	129.146
3.1822	3.0	237	3.0756	0.0	0.0	125.971
3.0081	4.0	316	2.9370	0.0	0.0	122.952
2.8946	5.0	395	2.8457	0.0	0.0	122.085
2.8162	6.0	474	2.7805	0.0000	0.0	121.204
2.7578	7.0	553	2.7284	0.0000	0.0	120.8485
2.7107	8.0	632	2.6850	0.0	0.0	120.5575
2.6695	9.0	711	2.6455	0.0000	0.0	120.3835
2.632	10.0	790	2.6074	0.0000	0.0	120.0615
2.5971	11.0	869	2.5695	0.0001	0.0	117.5055
2.5649	12.0	948	2.5360	0.0002	0.0	108.9205
2.5353	13.0	1027	2.5029	0.0003	0.0	95.6955
2.506	14.0	1106	2.4734	0.0004	0.0	85.586
2.4807	15.0	1185	2.4449	0.0005	0.0	77.844
2.455	16.0	1264	2.4136	0.0011	0.0	73.5575
2.4185	17.0	1343	2.3630	0.0013	0.0	68.8565
2.371	18.0	1422	2.2994	0.0024	0.0	64.448
2.3213	19.0	1501	2.2370	0.0027	0.0	62.501
2.2707	20.0	1580	2.1751	0.0039	0.0	59.419
2.227	21.0	1659	2.1183	0.0037	0.0	58.6405
2.182	22.0	1738	2.0610	0.0043	0.0	56.788
2.1396	23.0	1817	2.0002	0.0044	0.0	55.4195
2.0969	24.0	1896	1.9433	0.0046	0.0	54.239
2.0581	25.0	1975	1.8935	0.0046	0.0	52.833
2.025	26.0	2054	1.8459	0.0037	0.0	51.9935
1.9885	27.0	2133	1.7941	0.0043	0.0	50.6845
1.9587	28.0	2212	1.7568	0.0042	0.0	49.62
1.93	29.0	2291	1.7101	0.0047	0.0	48.6285
1.8983	30.0	2370	1.6641	0.0050	0.0	47.612
1.8693	31.0	2449	1.6341	0.0054	0.0	46.9725
1.8421	32.0	2528	1.5895	0.0049	0.0	46.026
1.8157	33.0	2607	1.5549	0.0057	0.0	45.169
1.7952	34.0	2686	1.5340	0.0058	0.0	44.602
1.7736	35.0	2765	1.4917	0.0065	0.0	43.823
1.7483	36.0	2844	1.4561	0.0055	0.0	43.098
1.7218	37.0	2923	1.4206	0.0071	0.0	42.265
1.6995	38.0	3002	1.3885	0.0065	0.0	41.419
1.6819	39.0	3081	1.3714	0.0057	0.0	41.078
1.6641	40.0	3160	1.3450	0.0066	0.0	40.324
1.6437	41.0	3239	1.3164	0.0053	0.0	39.8805
1.6198	42.0	3318	1.2894	0.0050	0.0	39.559
1.6045	43.0	3397	1.2686	0.0060	0.0	39.1475
1.5891	44.0	3476	1.2373	0.0069	0.0	38.3675
1.5774	45.0	3555	1.2252	0.0058	0.0	38.3125
1.5608	46.0	3634	1.2069	0.0056	0.0	38.1185
1.5488	47.0	3713	1.1713	0.0062	0.0	37.791
1.5265	48.0	3792	1.1443	0.0073	0.0	38.148
1.5095	49.0	3871	1.1223	0.0060	0.0	38.268
1.4939	50.0	3950	1.0998	0.0066	0.0	38.621
1.4799	51.0	4029	1.0816	0.0065	0.0	38.583

Framework versions

• Transformers 4.55.4
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4