deberta-v3-base-uner-down-synth400

prinsINT/deberta-v3-base-uner-down-synth400

deberta-v3-base-uner-down-synth400 is a token classification model from prinsINT. Use it when you need labels on individual words, such as names. It is set up for transformers. The card lists the license as mit.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

47% of all-time downloads

All-time downloads

Public

Parameters

184M

738 MB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors735 MB · 99%

At a glance

Task: Token Classification
Library: transformers
License: mit
Model type: deberta-v2
Access: Public
Created: Nov 26, 2025
Updated: Nov 26, 2025
SHA: 91168cd5

Try a prompt

Base models

microsoft/deberta-v3-base

Model card

From the Hugging Face model README

deberta-v3-base-uner-down-synth400

This model is a fine-tuned version of microsoft/deberta-v3-base on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.1225
F1: 0.7274
Precision: 0.6706
Recall: 0.7946
Accuracy: 0.9784

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: 2.5e-05
train_batch_size: 16
eval_batch_size: 32
seed: 42
optimizer: Use OptimizerNames.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: 30

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Precision	Recall	Accuracy
0.3736	0.8	20	0.2035	0.0482	0.1176	0.0303	0.9431
0.2416	1.6	40	0.1315	0.3044	0.2724	0.3449	0.9559
0.0684	2.4	60	0.1027	0.4766	0.4303	0.5341	0.9668
0.0588	3.2	80	0.0881	0.6067	0.5490	0.6778	0.9733
0.1194	4.0	100	0.0851	0.6805	0.6521	0.7114	0.9766
0.014	4.8	120	0.0800	0.7078	0.6512	0.7751	0.9769
0.0572	5.6	140	0.0815	0.7228	0.6882	0.7611	0.9788
0.0059	6.4	160	0.0910	0.7016	0.6408	0.7751	0.9767
0.0079	7.2	180	0.0946	0.6864	0.6237	0.7632	0.9760
0.0152	8.0	200	0.0981	0.7107	0.6494	0.7849	0.9760
0.0169	8.8	220	0.0954	0.702	0.6530	0.7589	0.9771
0.0027	9.6	240	0.0983	0.7214	0.6984	0.7459	0.9775
0.0107	10.4	260	0.1050	0.7141	0.6544	0.7859	0.9773
0.0029	11.2	280	0.1072	0.7139	0.6555	0.7838	0.9773
0.0056	12.0	300	0.1075	0.7216	0.6670	0.7859	0.9777
0.0048	12.8	320	0.1109	0.7245	0.6628	0.7989	0.9775
0.0033	13.6	340	0.1133	0.7242	0.6691	0.7892	0.9773
0.0021	14.4	360	0.1098	0.7247	0.6916	0.7611	0.9784
0.0057	15.2	380	0.1131	0.7223	0.6652	0.7903	0.9779
0.0014	16.0	400	0.1113	0.7319	0.6889	0.7805	0.9789
0.0098	16.8	420	0.1140	0.7207	0.6670	0.7838	0.9778
0.0019	17.6	440	0.1162	0.7177	0.6612	0.7849	0.9778
0.001	18.4	460	0.1194	0.7259	0.6682	0.7946	0.9780
0.0009	19.2	480	0.1174	0.7307	0.6860	0.7816	0.9784
0.001	20.0	500	0.1219	0.7267	0.6673	0.7978	0.9779
0.0016	20.8	520	0.1183	0.7312	0.6819	0.7881	0.9784
0.001	21.6	540	0.1187	0.7306	0.6850	0.7827	0.9784
0.0012	22.4	560	0.1227	0.7323	0.6752	0.8	0.9784
0.0011	23.2	580	0.1218	0.7212	0.6624	0.7914	0.9777
0.0012	24.0	600	0.1217	0.7243	0.6700	0.7881	0.9780
0.0006	24.8	620	0.1217	0.7296	0.6807	0.7859	0.9784
0.0006	25.6	640	0.1233	0.7272	0.6688	0.7968	0.9782
0.001	26.4	660	0.1206	0.7305	0.6807	0.7881	0.9787
0.0005	27.2	680	0.1207	0.7329	0.6842	0.7892	0.9786
0.0012	28.0	700	0.1211	0.7318	0.6822	0.7892	0.9786
0.0017	28.8	720	0.1225	0.7274	0.6706	0.7946	0.9784
0.001	29.6	740	0.1225	0.7274	0.6706	0.7946	0.9784

Framework versions

Transformers 4.57.1
Pytorch 2.8.0+cu128
Datasets 4.3.0
Tokenizers 0.22.1

Task: Token Classification
Library: transformers
Type: deberta-v2
License: mit
Created: Nov 26, 2025
Updated: Nov 26, 2025