canary-speech-translation-coreml / prompt_tokens.json
alexwengg's picture
Speech translation usage + FLEURS benchmarks for canary-1b-v2-coreml
ccceb4c verified
Raw
History Blame Contribute Delete
762 Bytes
{
"prompt_layout": [
"\u2581(16053)",
"<|startofcontext|>(7)",
"<|startoftranscript|>(4)",
"<|emo:undefined|>(16)",
"<SOURCE_LANG>",
"<TARGET_LANG>",
"<|pnc|>(5)/<|nopnc|>(6)",
"<|noitn|>(9)",
"<|notimestamp|>(11)",
"<|nodiarize|>(13)"
],
"task_rule": "transcribe when SOURCE==TARGET, translate when they differ",
"language_token_ids": {
"bg": 46,
"hr": 58,
"cs": 59,
"da": 60,
"nl": 62,
"en": 64,
"et": 66,
"fi": 70,
"fr": 71,
"de": 78,
"el": 79,
"hu": 89,
"it": 99,
"lv": 117,
"lt": 120,
"mt": 127,
"pl": 150,
"pt": 151,
"ro": 154,
"ru": 157,
"sk": 167,
"sl": 168,
"es": 171,
"sv": 175,
"uk": 192
},
"example_en_to_de": [
16053,
7,
4,
16,
64,
78,
5,
9,
11,
13
]
}