@juspay/neurolink
Version:
The pipe layer of an AI nervous system: one interface connecting provider neurons to your application, across three inference types — generate, stream and decide. `decide` returns typed, calibrated judgements from a non-generative model (~400ms, ~$0.00002
43 lines (42 loc) • 1.5 kB
JavaScript
/**
* TypeSafe (Jev) — the `decide` inference type, not text generation.
*
* `contextWindow` is genuinely meaningful here: it is the budget for the
* `state` being judged, measured by bisection at ~33,000 tokens for state
* plus the longest single question. (A separate, larger ~64,000 ceiling
* governs state plus ALL questions; the manifest has nowhere to express a
* second limit, and the tighter one is the safe number to publish.)
*
* `maxOutputTokens` is required by the manifest type but has no honest value
* for a model that emits no text — Jev returns a handful of numbers per
* question, so a small non-zero figure is the least misleading choice.
* `functionCalling` and `vision` are false: it has neither.
*/
export const typesafeManifest = {
defaultContextWindow: 33000,
models: {
_default: {
aliases: ["jev"],
contextWindow: 33000,
maxOutputTokens: 1024,
vision: false,
functionCalling: false,
},
"jev-latest": {
aliases: ["jev"],
displayName: "Jev (latest)",
contextWindow: 33000,
maxOutputTokens: 1024,
vision: false,
functionCalling: false,
},
"jev-preview": {
aliases: [],
displayName: "Jev (preview)",
contextWindow: 33000,
maxOutputTokens: 1024,
vision: false,
functionCalling: false,
},
},
};