j]TdZddlmZddlZgdZehdZdddd Zd d Zd d Z y)aRecover a model's Hub id from the name the results bucket uses. A manifest filename cannot carry the ``/`` of a Hub id, so the eval scripts write ``model_id.replace("/", "-")`` (see :func:`normalizer.eval_utils`), and the bucket directory is named the same way. That transform is lossy: once the separator is a hyphen, nothing in ``abr-ai-niagara-19m-batch.en`` says whether the org is ``abr`` or ``abr-ai``. The tables below carry the information the filename dropped. >>> to_hub_id("microsoft-Phi-4-multimodal-instruct") 'microsoft/Phi-4-multimodal-instruct' >>> to_hub_id("ibm-granite-granite-speech-3.3-8b") 'ibm-granite/granite-speech-3.3-8b' ) annotationsN)HYPHENATED_ORGSBUCKET_ID_OVERRIDES to_hub_id to_hub_ids>abr-aizai-org AutoArk-AI ibm-granite ARTPARK-IISc OpenMOSS-Teamdistil-whisperefficient-speechzfacebook/omniASR-CTC-7B-v2zfacebook/omniASR-LLM-7B-v2z(nvidia/stt_en_conformer_transducer_small)omniASR_CTC_7B_v2omniASR_LLM_7B_v2!stt_en_conformer_transducer_smallc|tvr t|StttdD].}|j |ds|d|t|dzdcS|j d\}}}|r|d|S|S)a1The Hub id ``bucket_id`` was written from, e.g. ``microsoft/Phi-4-multimodal-instruct``. Returns ``bucket_id`` unchanged when it carries no separator to restore and no override claims it, so an unrecognised name passes through rather than being mangled into a plausible-looking wrong id. T)keyreverse-/N)rsortedrlen startswith partition) bucket_idorgsepnames /app/normalizer/model_ids.pyrr2s''"9--o3=8   3%q *U!Ic#hln567 78((-NCd!cU!D6?0y0c|Dcic]}|t|}}tj|j}t d|j D}|r[|D]Et fd|j D}t ddj|ddG|Dcic]}||c}S|Scc}wcc}w)aMap each bucket model name in ``models`` to the Hub id to report it under. :func:`to_hub_id` applied one at a time cannot see that two bucket names landed on the same Hub id, which would silently merge two models' rows into one. Checked here instead: on a collision, report it and fall back to the bucket names for the whole set, since a recognisable wrong name beats a plausible-looking wrong number. c32K|]\}}|dkDs |yw)rN).0hubns r! zto_hub_ids..OsDfc1a!ecDs c34K|]\}}|k(s |yw)Nr%)r&mhr's r!r)zto_hub_ids..RsC41a!s(QCs z warning: z and z both map to z; reporting bucket names)r collectionsCountervaluesritemsprintjoin)modelsmodelout collisionsclashingsourcesr's @r!rrDs17 7u5)E" " 7C 7$$SZZ\2JD (8(8(:DDH cCC399;CCG K W 56mC5H`a b c+11u 11 J 82s B=. C)rstrreturnr9)r3z list[str]r:zdict[str, str]) __doc__ __future__rr-__all__ frozensetrrrrr%r"r!r?sE # O   &65)S1$r"