{ "title": "Zenthi-AI Milestone 1: Data Refinery & Decontamination Report", "total_train_samples": 15000, "total_eval_samples": 600, "decontaminated_humaneval_samples_purged": 3514, "quality_auditor_rejected_samples": 4028, "domain_breakdown": { "code": { "train_count": 4000, "eval_count": 100, "total_curated": 4100 }, "ocr_vision": { "train_count": 3000, "eval_count": 100, "total_curated": 3100 }, "web_extraction": { "train_count": 3000, "eval_count": 100, "total_curated": 3100 }, "reasoning": { "train_count": 2000, "eval_count": 100, "total_curated": 2100 }, "doc_generation": { "train_count": 2000, "eval_count": 100, "total_curated": 2100 }, "diagram_analysis": { "train_count": 1000, "eval_count": 100, "total_curated": 1100 } }, "train_file": "F:\\Zenith-AI\\datasets\\zenthi_hybrid_train.jsonl", "eval_file": "F:\\Zenith-AI\\datasets\\zenthi_hybrid_eval.jsonl" }