aman/benchmarks/heuristics_dataset.raw.jsonl
Thales Maciel 8c1f7c1e13
Some checks failed
ci / test-and-build (push) Has been cancelled
Add benchmark-driven model promotion workflow and pipeline stages
2026-02-28 15:12:33 -03:00

8 lines
2.3 KiB
JSON

{"id":"corr-time-01","transcript":"set alarm for 6 i mean 7","words":[{"text":"set","start_s":0.0,"end_s":0.1,"prob":0.9},{"text":"alarm","start_s":0.2,"end_s":0.3,"prob":0.9},{"text":"for","start_s":0.4,"end_s":0.5,"prob":0.9},{"text":"6","start_s":0.6,"end_s":0.7,"prob":0.9},{"text":"i","start_s":0.8,"end_s":0.9,"prob":0.9},{"text":"mean","start_s":1.0,"end_s":1.1,"prob":0.9},{"text":"7","start_s":1.2,"end_s":1.3,"prob":0.9}],"expected_aligned_text":"set alarm for 7","expected":{"applied_min":1,"required_rule_ids":["cue_correction"]},"tags":["i_mean_correction","timing_sensitive"]}
{"id":"corr-time-gap-01","transcript":"set alarm for 6 i mean 7","words":[{"text":"set","start_s":0.0,"end_s":0.1,"prob":0.9},{"text":"alarm","start_s":0.2,"end_s":0.3,"prob":0.9},{"text":"for","start_s":0.4,"end_s":0.5,"prob":0.9},{"text":"6","start_s":0.6,"end_s":0.7,"prob":0.9},{"text":"i","start_s":2.0,"end_s":2.1,"prob":0.9},{"text":"mean","start_s":2.2,"end_s":2.3,"prob":0.9},{"text":"7","start_s":2.4,"end_s":2.5,"prob":0.9}],"expected_aligned_text":"set alarm for 6 i mean 7","expected":{"applied_min":0,"forbidden_rule_ids":["cue_correction"]},"tags":["i_mean_literal","timing_sensitive"]}
{"id":"literal-mean-01","transcript":"write exactly i mean this sincerely","expected_aligned_text":"write exactly i mean this sincerely","expected":{"applied_min":0,"forbidden_rule_ids":["cue_correction"]},"tags":["i_mean_literal"]}
{"id":"restart-01","transcript":"please send it please send it","expected_aligned_text":"please send it","expected":{"applied_min":1,"required_rule_ids":["restart_repeat"]},"tags":["restart"]}
{"id":"actually-correction-01","transcript":"set alarm for 6 actually 7","expected_aligned_text":"set alarm for 7","expected":{"applied_min":1,"required_rule_ids":["cue_correction"]},"tags":["actually_correction"]}
{"id":"sorry-correction-01","transcript":"set alarm for 6 sorry 7","expected_aligned_text":"set alarm for 7","expected":{"applied_min":1,"required_rule_ids":["cue_correction"]},"tags":["sorry_correction"]}
{"id":"no-correction-phrase-01","transcript":"set alarm for 6 i mean","expected_aligned_text":"set alarm for 6 i mean","expected":{"applied_min":0,"forbidden_rule_ids":["cue_correction"]},"tags":["i_mean_literal"]}
{"id":"baseline-unchanged-01","transcript":"hello world","expected_aligned_text":"hello world","expected":{"applied_min":0},"tags":["baseline"]}