bert-base-stage2-sbert

SentenceTransformer checkpoint fine-tuned for Vietnamese legal retrieval.

Evaluation

• Truncate dims: [768, 512, 256, 128]

UIT-ViQuAD2.0

method	Accuracy@1	Accuracy@3	Accuracy@5	Accuracy@10	NDCG@3	NDCG@5	NDCG@10	MRR@3	MRR@5	MRR@10	Recall@5	Recall@10	Recall@100	MAP@100
BM25	0.606492	0.75565	0.805643	0.861663	0.694646	0.715313	0.733532	0.673492	0.685004	0.692583	0.805643	0.861663	0.966717	0.69741
768	0.498288	0.676209	0.748665	0.824819	0.603137	0.63304	0.657852	0.57782	0.594448	0.604798	0.748665	0.824819	0.964662	0.611125
512	0.49007	0.669086	0.742501	0.818929	0.595305	0.625579	0.650532	0.569762	0.586582	0.59702	0.742501	0.818929	0.96343	0.603549
256	0.472127	0.648952	0.724969	0.804547	0.576285	0.607607	0.633454	0.551112	0.5685	0.579232	0.724969	0.804547	0.957951	0.586068
128	0.440077	0.626216	0.695795	0.778387	0.549035	0.577717	0.604581	0.522349	0.538278	0.549453	0.695795	0.778387	0.947815	0.556887

Zalo-Legal

method	Accuracy@1	Accuracy@3	Accuracy@5	Accuracy@10	NDCG@3	NDCG@5	NDCG@10	MRR@3	MRR@5	MRR@10	Recall@5	Recall@10	Recall@100	MAP@100
BM25	0.379442	0.616751	0.69797	0.77665	0.515637	0.549114	0.57497	0.48181	0.500402	0.511188	0.696701	0.775381	0.923858	0.517283
768	0.369289	0.557107	0.651015	0.782995	0.47751	0.515787	0.559141	0.451988	0.47261	0.490666	0.649112	0.781726	0.961929	0.498361
512	0.371827	0.538071	0.630711	0.769036	0.467101	0.50561	0.550864	0.444585	0.465778	0.484756	0.628807	0.767132	0.956853	0.493436
256	0.351523	0.525381	0.611675	0.765228	0.451601	0.487877	0.537285	0.428088	0.448266	0.468514	0.609772	0.763325	0.951777	0.476734
128	0.326142	0.491117	0.56599	0.729695	0.421049	0.45224	0.50528	0.3989	0.416032	0.437983	0.564086	0.727792	0.937817	0.447903