benchmark_report.md

Vaidya Benchmark Report

Stage 1: Parser

• total_verses: 8741
• citation_completeness: 1.0
• sthana_distribution: {'Sutrasthana': 1710, 'Vimanasthana': 341, 'Siddhisthana': 589, 'Nidanasthana': 277, 'Sharirasthana': 763, 'Cikitsasthana': 4756, 'Kalpasthana': 305}
• avg_verse_length_chars: 99.74
• empty_or_short_verses: 4
• devanagari_ratio: 0.8923
• PASS: True

Stage 2: Chunker

• total_chunks: 1612
• avg_chunk_tokens: 222.07
• citation_present_ratio: 1.0
• semantic_boundary_broken_ratio: 0.0
• outlier_chunks: 0
• PASS: True

Stage 3: Translator

• total_translated_chunks: 1612
• empty_english_ratio: 0.0
• empty_hindi_ratio: 0.0
• keyword_presence_ratio: 0.65
• PASS: True

Stage 4: Embedder

• total_vectors: 1612
• vector_dimension: 1024
• collection_name: vaidya_caraka
• retrieval_functional: True
• PASS: True