{ "artifact": "tokenizer/development_variants/opcode_5_250.json", "corpus_scope": "official-train", "token_field": "opcodes", "processed_root": "data/processed_variants/filter_5_250", "train_csv": [ "data/splits/seed42/train.csv", "data/splits/seed42/valid.csv" ], "excluded_csv": "data/excluded_samples.csv", "manifest": "data/manifest.csv", "active_sample_count": 2123, "function_count": 962505, "requested_vocab_size": 8192, "actual_vocab_size": 978, "min_frequency": 2, "normalization_schema": "asm-operand-v3-fp-pointer", "variant_protocol": "data/processed_variants/filter_5_250/variant_protocol.json", "environment_profile": "paper-env-v2", "environment_lock_sha256": "4b0c5decdac180d4b220e1aef5bba344833d885146c63da1fb08afec4126b311" }