TinyQuery-140M / tokenization.json
karmx's picture
Release TinyQuery 139.7M from scratch with frozen weights, reproducible Mac evaluations and runtime source
b296ad4 verified
Raw
History Blame Contribute Delete
992 Bytes
{
"vocab_size": 4082,
"context": 2048,
"special_tokens": {
"<|pad|>": 0,
"<|context|>": 1,
"<|user|>": 2,
"<|assistant|>": 3,
"<|end|>": 4
},
"train": {
"examples": 347376,
"tokens": 100224712,
"response_tokens": 13163249,
"length_p50": 278.0,
"length_p95": 389.0,
"max_length": 945,
"rejected": [],
"ids_sha256": "365f03740dee120064b98b530f4b67e987f1cf192435281bfd9d3f32f4271f75"
},
"validation": {
"examples": 1200,
"tokens": 336514,
"response_tokens": 42286,
"length_p50": 276.0,
"length_p95": 374.0,
"max_length": 406,
"rejected": [],
"ids_sha256": "7bdf2a15c82cd7459e2410cd3fd7cf70410392780e9979228b1edaec1914e4bf"
},
"test": {
"examples": 1196,
"tokens": 339535,
"response_tokens": 43042,
"length_p50": 279.0,
"length_p95": 388.0,
"max_length": 426,
"rejected": [],
"ids_sha256": "1c668ade4e6eff7d1e62697b39419ca59befc572f6961802cad0570e15166d41"
}
}