SmolVLM-Base-ocr-isl-with-isl-backbone

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

• Loss: 0.0147
• Wer: 0.2907
• Cer: 0.5314
• Exact Match: 0.0
• Special Char Acc: 1.0
• Seq Acc 5: 0.0
• Seq Acc 10: 0.0

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: 0.0003
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Exact Match	Special Char Acc	Seq Acc 5	Seq Acc 10
0.3666	0.0325	125	0.2243	0.8704	0.8629	0.0	0.9622	0.0	0.0
0.168	0.0649	250	0.1485	0.7556	0.9306	0.0	0.9680	0.0	0.0
0.1282	0.0974	375	0.1187	0.4481	0.6394	0.0	0.9797	0.0	0.0
0.0984	0.1299	500	0.0965	0.5056	0.7014	0.0	0.9826	0.0	0.0
0.0891	0.1624	625	0.0755	0.4611	0.6485	0.0	0.9913	0.0	0.0
0.0744	0.1948	750	0.0638	0.4963	0.7116	0.0	0.9913	0.0	0.0
0.0708	0.2273	875	0.0518	0.3944	0.5805	0.0	0.9942	0.0	0.0
0.0647	0.2598	1000	0.0611	0.5389	0.8122	0.0	0.9855	0.0	0.0
0.0572	0.2922	1125	0.0454	0.4796	0.7158	0.0	0.9913	0.0	0.0
0.0555	0.3247	1250	0.0320	0.5685	0.7432	0.0	0.9884	0.0	0.0
0.0445	0.3572	1375	0.0386	0.4611	0.6404	0.0	0.9971	0.0	0.0
0.0455	0.3897	1500	0.0392	0.4259	0.6783	0.0	0.9913	0.0	0.0
0.0469	0.4221	1625	0.0319	0.2944	0.6415	0.0	0.9971	0.0	0.0
0.0386	0.4546	1750	0.0305	0.3574	0.5656	0.0	0.9971	0.0	0.0
0.0393	0.4871	1875	0.0327	0.2889	0.5405	0.0	0.9971	0.0	0.0
0.0364	0.5195	2000	0.0254	0.2685	0.4360	0.0	0.9971	0.0	0.0
0.0338	0.5520	2125	0.0254	0.2556	0.4616	0.0	1.0	0.0	0.0
0.0332	0.5845	2250	0.0217	0.3111	0.5105	0.0	1.0	0.0	0.0
0.0285	0.6170	2375	0.0257	0.3167	0.4898	0.0	1.0	0.0	0.0
0.0291	0.6494	2500	0.0230	0.4481	0.6054	0.0	1.0	0.0	0.0
0.028	0.6819	2625	0.0204	0.3741	0.5687	0.0	1.0	0.0	0.0
0.024	0.7144	2750	0.0204	0.3352	0.5242	0.0	0.9971	0.0	0.0
0.0262	0.7469	2875	0.0170	0.3796	0.6020	0.0	1.0	0.0	0.0
0.0237	0.7793	3000	0.0155	0.4222	0.6574	0.0	1.0	0.0	0.0
0.0246	0.8118	3125	0.0164	0.2796	0.5151	0.0	1.0	0.0	0.0
0.023	0.8443	3250	0.0152	0.2815	0.5054	0.0	1.0	0.0	0.0
0.0225	0.8767	3375	0.0150	0.2722	0.5151	0.0	1.0	0.0	0.0
0.0216	0.9092	3500	0.0153	0.2815	0.5268	0.0	1.0	0.0	0.0
0.0213	0.9417	3625	0.0149	0.2852	0.5300	0.0	1.0	0.0	0.0
0.0204	0.9742	3750	0.0147	0.2907	0.5314	0.0	1.0	0.0	0.0

Framework versions

• PEFT 0.18.0
• Transformers 4.57.3
• Pytorch 2.9.1+cu128
• Datasets 4.4.1
• Tokenizers 0.22.1