dchen0/font-classifier

Merged DINOv2‑base checkpoint with LoRA weights for font classification.

This model is a fine-tuned version of facebook/dinov2-base-imagenet1k-1-layer on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.2637
• Model Preparation Time: 0.0016
• Accuracy: 0.9163

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.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: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Accuracy
0.7099	0.0182	50	0.6595	0.0016	0.7594
0.7084	0.0363	100	0.6175	0.0016	0.7806
0.7638	0.0545	150	0.7014	0.0016	0.7337
0.6451	0.0727	200	0.6177	0.0016	0.7757
0.6852	0.0908	250	0.5691	0.0016	0.7971
0.5753	0.1090	300	0.5666	0.0016	0.8048
0.5925	0.1272	350	0.5235	0.0016	0.8204
0.6969	0.1453	400	0.5725	0.0016	0.7922
0.6096	0.1635	450	0.5103	0.0016	0.8173
0.5994	0.1817	500	0.5075	0.0016	0.8183
0.5272	0.1999	550	0.5116	0.0016	0.8229
0.5193	0.2180	600	0.4952	0.0016	0.8244
0.5689	0.2362	650	0.4662	0.0016	0.8388
0.5126	0.2544	700	0.4651	0.0016	0.8327
0.5301	0.2725	750	0.5080	0.0016	0.8158
0.5424	0.2907	800	0.4573	0.0016	0.8357
0.4357	0.3089	850	0.4412	0.0016	0.8486
0.5522	0.3270	900	0.4755	0.0016	0.8256
0.5639	0.3452	950	0.4463	0.0016	0.8339
0.4522	0.3634	1000	0.4347	0.0016	0.8458
0.5548	0.3815	1050	0.4112	0.0016	0.8560
0.4815	0.3997	1100	0.4300	0.0016	0.8514
0.5028	0.4179	1150	0.3840	0.0016	0.8713
0.4417	0.4360	1200	0.4364	0.0016	0.8462
0.4465	0.4542	1250	0.3731	0.0016	0.8740
0.3935	0.4724	1300	0.3672	0.0016	0.8753
0.5306	0.4906	1350	0.4480	0.0016	0.8388
0.3991	0.5087	1400	0.3718	0.0016	0.8698
0.483	0.5269	1450	0.3916	0.0016	0.8652
0.4323	0.5451	1500	0.3948	0.0016	0.8648
0.3664	0.5632	1550	0.3400	0.0016	0.8796
0.4941	0.5814	1600	0.3531	0.0016	0.8765
0.4185	0.5996	1650	0.3481	0.0016	0.8820
0.4506	0.6177	1700	0.3332	0.0016	0.8866
0.4015	0.6359	1750	0.3468	0.0016	0.8768
0.3919	0.6541	1800	0.3421	0.0016	0.8897
0.4281	0.6722	1850	0.3141	0.0016	0.8937
0.3659	0.6904	1900	0.3424	0.0016	0.8823
0.345	0.7086	1950	0.3172	0.0016	0.8912
0.3157	0.7267	2000	0.3226	0.0016	0.8903
0.3456	0.7449	2050	0.3178	0.0016	0.8909
0.3643	0.7631	2100	0.2988	0.0016	0.8983
0.4043	0.7812	2150	0.3036	0.0016	0.8992
0.3486	0.7994	2200	0.2974	0.0016	0.9053
0.3735	0.8176	2250	0.3026	0.0016	0.8964
0.4032	0.8358	2300	0.2990	0.0016	0.9019
0.3825	0.8539	2350	0.2938	0.0016	0.9062
0.345	0.8721	2400	0.2871	0.0016	0.9059
0.3528	0.8903	2450	0.2777	0.0016	0.9093
0.3207	0.9084	2500	0.2764	0.0016	0.9111
0.2664	0.9266	2550	0.2741	0.0016	0.9099
0.3496	0.9448	2600	0.2720	0.0016	0.9151
0.3274	0.9629	2650	0.2724	0.0016	0.9136
0.3014	0.9811	2700	0.2659	0.0016	0.9136
0.3235	0.9993	2750	0.2637	0.0016	0.9163

Framework versions

• PEFT 0.15.2
• Transformers 4.52.4
• Pytorch 2.7.1
• Datasets 3.6.0
• Tokenizers 0.21.1