Camelot pipeline with LLM fallback achieves highest accuracy (EM/LS up to 0.99-1.00) and efficiency (<1s per PDF) compared to LLM-only or hybrid regex-LLM on 860 documents using local 12-14B models.
Hubungan Penerapan Sistem Blok UNIMED terhadap Efektivitas Pembelajaran Statistik Pendidikan Mahasiswa
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Tabular PDF Information Extraction with Local LLMs and Layout-Aware Parsing: A Reliability Evaluation
Camelot pipeline with LLM fallback achieves highest accuracy (EM/LS up to 0.99-1.00) and efficiency (<1s per PDF) compared to LLM-only or hybrid regex-LLM on 860 documents using local 12-14B models.