We benchmark parameter efficiency, memory latency, and semantic retrieval accuracy across 15 transformer architectures evaluated on the CrossRef and Google Scholar metadata corpora. Empirical results demonstrate a 24% reduction in inference overhead with minimal recall degradation.
AIT: AIT-2026-00043
JIT: JIT-RESPO-02
DOI: 10.5555/respo.2026.00043
IIT: IIT-MIT-09 Artificial Intelligence
Transformers vs. Large Foundation Models: Benchmarking Empirical Complexity in Scholarly Information Retrieval
Authors & Affiliations (1 Contributor)
Prof. Elena Rostova 1st Author
Principal Research Scientist • Computer Science and Artificial Intelligence Laboratory (MIT CSAIL)
Journal: RESPO Journal of Computational Science & Artificial Intelligence •
ISSN: 2990-1234 •
Vol. 4, Issue 2 (2026)
Received: Oct 4, 2026 | Published: 2026-10-04 07:35:01
Article Views
3,855
PDF Downloads
1,292
Citations
92
Article Index Rank
Q1 (Top 5%)
Article Impact Value
4.85
Journal Impact
6.42
Abstract
We benchmark parameter efficiency, memory latency, and semantic retrieval accuracy across 15 transformer architectures evaluated on the CrossRef and Google Scholar metadata corpora. Empirical results demonstrate a 24% reduction in inference overhead with minimal recall degradation.
Keywords:
transformers, foundation models, information retrieval, latent semantic vectors
Full Manuscript Content
Cite this Article (BibTeX Format)
Strictly formatted for LaTeX, Zotero, Mendeley, and Google Scholar bibliography import.
@article{admin2026_28,
title = {Transformers vs. Large Foundation Models: Benchmarking Empirical Complexity in Scholarly Information Retrieval},
author = {admin},
journal = {RESPO Journal of Computational Science & Artificial Intelligence},
volume = {1},
number = {1},
pages = {1--10},
year = {2026},
doi = {10.5555/respo.2026.00043},
note = {AIT: AIT-2026-00043},
url = {https://respo.iledu.in/articles/transformers-vs-large-foundation-models-benchmarking-empirical-complexity-in-scholarly-information-retrieval/}
}