Interface: ToTOptions

Defined in: src/patterns/ToT.ts:26

Properties

beamWidth?

readonly optional beamWidth?: number

Defined in: src/patterns/ToT.ts:42

Beam width — how many thoughts survive after each level. Default 1 (greedy).

---

branchingFactor

readonly branchingFactor: number

Defined in: src/patterns/ToT.ts:34

Branching factor — K thoughts generated per frontier node per iteration.

---

depth

readonly depth: number

Defined in: src/patterns/ToT.ts:32

Depth of the tree (number of expansion iterations).

---

id?

readonly optional id?: string

Defined in: src/patterns/ToT.ts:46

---

maxTokens?

readonly optional maxTokens?: number

Defined in: src/patterns/ToT.ts:44

---

model

readonly model: string

Defined in: src/patterns/ToT.ts:28

---

name?

readonly optional name?: string

Defined in: src/patterns/ToT.ts:45

---

provider

readonly provider: LLMProvider

Defined in: src/patterns/ToT.ts:27

---

score

readonly score: (thought) => number

Defined in: src/patterns/ToT.ts:40

Scorer: given a thought, return a numeric score. Higher is better. The top beamWidth thoughts survive each level; the rest are pruned. Synchronous so pruning is deterministic.

Parameters

thought

string

Returns

number

---

temperature?

readonly optional temperature?: number

Defined in: src/patterns/ToT.ts:43

---

thoughtPrompt

readonly thoughtPrompt: string

Defined in: src/patterns/ToT.ts:30

System prompt for the thought-generation LLMCall.