@howells/motif-sdk 0.9.0 → 0.10.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +28 -0
- package/dist/image.d.ts +51 -1
- package/dist/image.js +85 -1
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -172,6 +172,34 @@ Four providers are implemented, each reading its own API key from the environmen
|
|
|
172
172
|
|
|
173
173
|
`generate()` and `edit()` accept `tier` (`"fast" | "balanced" | "quality" | "hero"`) to resolve a model per provider, or an explicit `model` id. Every result carries a normalized per-call `cost: { usd, source }`.
|
|
174
174
|
|
|
175
|
+
### Best-of-N with an injectable judge
|
|
176
|
+
|
|
177
|
+
`bestOfN()` generates `n` candidates in parallel and picks a winner. It reuses the same options as `generate()` (text→image) or `edit()` (pass `images` for the edit path), plus `n` and an optional `judge`. When a `seed` is given each candidate uses `seed + index`, so the N vary. The judge is a caller-provided function — the layer takes no text-client dependency, so it pairs well with `@howells/ai`'s vision client but does not require it. Omit the judge and candidate 0 wins.
|
|
178
|
+
|
|
179
|
+
```ts
|
|
180
|
+
const best = await img.bestOfN({
|
|
181
|
+
prompt: "a plain room, bare concrete wall",
|
|
182
|
+
n: 4,
|
|
183
|
+
seed: 100, // candidates get seeds 100, 101, 102, 103
|
|
184
|
+
// Caller-provided judge: receives the successful candidates + context,
|
|
185
|
+
// returns the winning index. Wire in @howells/ai here if you want a vision judge.
|
|
186
|
+
judge: async (candidates, context) => {
|
|
187
|
+
// ...score candidates[i].images[0] against context.prompt...
|
|
188
|
+
return { index: 0, reason: "sharpest wall texture" };
|
|
189
|
+
},
|
|
190
|
+
});
|
|
191
|
+
|
|
192
|
+
if (best.isOk()) {
|
|
193
|
+
best.value.best; // the winning MotifImageResult
|
|
194
|
+
best.value.chosenIndex; // its index within candidates
|
|
195
|
+
best.value.reason; // the judge's rationale, if any
|
|
196
|
+
best.value.candidates; // every successful candidate (generation order)
|
|
197
|
+
best.value.totalCostUsd; // summed USD across all candidates generated
|
|
198
|
+
}
|
|
199
|
+
```
|
|
200
|
+
|
|
201
|
+
If some candidates fail, the judge sees only the survivors; if all `n` fail, `bestOfN()` returns `Result.err`.
|
|
202
|
+
|
|
175
203
|
## Testing
|
|
176
204
|
|
|
177
205
|
```bash
|
package/dist/image.d.ts
CHANGED
|
@@ -166,12 +166,62 @@ interface MotifImageResult {
|
|
|
166
166
|
*/
|
|
167
167
|
warnings?: readonly string[];
|
|
168
168
|
}
|
|
169
|
+
/**
|
|
170
|
+
* Picks the winning candidate for a {@link MotifImageClient.bestOfN} call.
|
|
171
|
+
*
|
|
172
|
+
* Receives the successful candidates (index-aligned to
|
|
173
|
+
* {@link BestOfNResult.candidates}) and the request context, and returns the
|
|
174
|
+
* chosen index (plus an optional human-readable reason). May be sync or async.
|
|
175
|
+
* The judge is caller-provided so the image layer stays decoupled from any text
|
|
176
|
+
* client — it pairs well with `@howells/ai`'s vision client, but that dependency
|
|
177
|
+
* is not required.
|
|
178
|
+
*/
|
|
179
|
+
type ImageJudge = (candidates: readonly MotifImageResult[], context: {
|
|
180
|
+
readonly prompt?: string;
|
|
181
|
+
readonly instruction?: string;
|
|
182
|
+
}) => Promise<{
|
|
183
|
+
index: number;
|
|
184
|
+
reason?: string;
|
|
185
|
+
}> | {
|
|
186
|
+
index: number;
|
|
187
|
+
reason?: string;
|
|
188
|
+
};
|
|
189
|
+
/**
|
|
190
|
+
* Options for a best-of-N generation. Extends either {@link GenerateImageOptions}
|
|
191
|
+
* (text→image) or {@link EditImageOptions} (multi-image edit) — the presence of
|
|
192
|
+
* `images` selects the edit path — with the candidate count and an optional judge.
|
|
193
|
+
*/
|
|
194
|
+
type BestOfNOptions = (GenerateImageOptions | EditImageOptions) & {
|
|
195
|
+
/** How many candidates to generate (>= 1). */
|
|
196
|
+
n: number;
|
|
197
|
+
/** Chooses the winner. Omit → candidate 0 wins. */
|
|
198
|
+
judge?: ImageJudge;
|
|
199
|
+
};
|
|
200
|
+
/** Result of a {@link MotifImageClient.bestOfN} call. */
|
|
201
|
+
interface BestOfNResult {
|
|
202
|
+
/** The winning candidate (its single-image {@link MotifImageResult}). */
|
|
203
|
+
best: MotifImageResult;
|
|
204
|
+
/** The index of `best` within `candidates`. */
|
|
205
|
+
chosenIndex: number;
|
|
206
|
+
/** Judge's rationale, if it returned one. */
|
|
207
|
+
reason?: string;
|
|
208
|
+
/** All successful candidates, in generation order. */
|
|
209
|
+
candidates: readonly MotifImageResult[];
|
|
210
|
+
/** Total USD across all candidates that were generated (successes only). */
|
|
211
|
+
totalCostUsd: number;
|
|
212
|
+
}
|
|
169
213
|
/** The provider-agnostic image client. Every method returns a Result — no throws. */
|
|
170
214
|
interface MotifImageClient {
|
|
171
215
|
/** Text→image generation. */
|
|
172
216
|
generate: (opts: GenerateImageOptions) => Promise<Result<MotifImageResult, MotifError>>;
|
|
173
217
|
/** Multi-image edit (with optional mask). */
|
|
174
218
|
edit: (opts: EditImageOptions) => Promise<Result<MotifImageResult, MotifError>>;
|
|
219
|
+
/**
|
|
220
|
+
* Generate N candidates (in parallel) and pick the best via an optional
|
|
221
|
+
* injectable judge. Discriminates generate vs edit by the presence of
|
|
222
|
+
* `images`. Returns the winner plus all successful candidates and total spend.
|
|
223
|
+
*/
|
|
224
|
+
bestOfN: (opts: BestOfNOptions) => Promise<Result<BestOfNResult, MotifError>>;
|
|
175
225
|
}
|
|
176
226
|
|
|
177
227
|
/**
|
|
@@ -373,4 +423,4 @@ declare function costForImages(provider: ImageProviderId, modelId: string, provi
|
|
|
373
423
|
*/
|
|
374
424
|
declare function createMotifImage(config?: MotifImageConfig, deps?: MotifImageDeps): MotifImageClient;
|
|
375
425
|
|
|
376
|
-
export { type EditImageOptions, FAL_API_KEY_ENV, FAL_TIER_MODELS, GOOGLE_API_KEY_ENV, GOOGLE_TIER_MODELS, type GenerateImageOptions, type ImageCost, type ImageCostSource, type ImageProviderAdapter, type ImageProviderId, type ImageTier, type MotifImageClient, type MotifImageConfig, type MotifImageFile, type MotifImageResult, OPENAI_API_KEY_ENV, OPENAI_TIER_MODELS, PROVIDERS, REPLICATE_API_KEY_ENV, REPLICATE_TIER_MODELS, costForImages, costFromProviderMetadata, createMotifImage, getProviderAdapter };
|
|
426
|
+
export { type BestOfNOptions, type BestOfNResult, type EditImageOptions, FAL_API_KEY_ENV, FAL_TIER_MODELS, GOOGLE_API_KEY_ENV, GOOGLE_TIER_MODELS, type GenerateImageOptions, type ImageCost, type ImageCostSource, type ImageJudge, type ImageProviderAdapter, type ImageProviderId, type ImageTier, type MotifImageClient, type MotifImageConfig, type MotifImageFile, type MotifImageResult, OPENAI_API_KEY_ENV, OPENAI_TIER_MODELS, PROVIDERS, REPLICATE_API_KEY_ENV, REPLICATE_TIER_MODELS, costForImages, costFromProviderMetadata, createMotifImage, getProviderAdapter };
|
package/dist/image.js
CHANGED
|
@@ -1534,7 +1534,91 @@ function createMotifImage(config = {}, deps = {}) {
|
|
|
1534
1534
|
return err2(toMotifError(error));
|
|
1535
1535
|
}
|
|
1536
1536
|
}
|
|
1537
|
-
|
|
1537
|
+
async function bestOfN(opts) {
|
|
1538
|
+
return await runBestOfN(opts, generate, edit);
|
|
1539
|
+
}
|
|
1540
|
+
return { generate, edit, bestOfN };
|
|
1541
|
+
}
|
|
1542
|
+
async function runCandidate(opts, index, generate, edit) {
|
|
1543
|
+
const candidateSeed = opts.seed === void 0 ? void 0 : opts.seed + index;
|
|
1544
|
+
const overrides = {
|
|
1545
|
+
n: 1,
|
|
1546
|
+
...candidateSeed === void 0 ? {} : { seed: candidateSeed }
|
|
1547
|
+
};
|
|
1548
|
+
if ("images" in opts) {
|
|
1549
|
+
return await edit({ ...opts, ...overrides });
|
|
1550
|
+
}
|
|
1551
|
+
return await generate({ ...opts, ...overrides });
|
|
1552
|
+
}
|
|
1553
|
+
async function selectWinner(opts, candidates) {
|
|
1554
|
+
const { judge } = opts;
|
|
1555
|
+
if (judge === void 0) {
|
|
1556
|
+
return ok2({ chosenIndex: 0 });
|
|
1557
|
+
}
|
|
1558
|
+
const context = "images" in opts ? { instruction: opts.instruction } : { prompt: opts.prompt };
|
|
1559
|
+
const decision = await judge(candidates, context);
|
|
1560
|
+
if (!Number.isInteger(decision.index) || decision.index < 0 || decision.index >= candidates.length) {
|
|
1561
|
+
return err2(
|
|
1562
|
+
new MotifError(
|
|
1563
|
+
`bestOfN judge returned an out-of-range index ${decision.index} (expected 0..${candidates.length - 1})`,
|
|
1564
|
+
0
|
|
1565
|
+
)
|
|
1566
|
+
);
|
|
1567
|
+
}
|
|
1568
|
+
return ok2({
|
|
1569
|
+
chosenIndex: decision.index,
|
|
1570
|
+
...decision.reason === void 0 ? {} : { reason: decision.reason }
|
|
1571
|
+
});
|
|
1572
|
+
}
|
|
1573
|
+
async function runBestOfN(opts, generate, edit) {
|
|
1574
|
+
try {
|
|
1575
|
+
const { n } = opts;
|
|
1576
|
+
if (!Number.isInteger(n) || n < 1) {
|
|
1577
|
+
return err2(
|
|
1578
|
+
new MotifError(`bestOfN requires an integer n >= 1 (got ${n})`, 0)
|
|
1579
|
+
);
|
|
1580
|
+
}
|
|
1581
|
+
const results = await Promise.all(
|
|
1582
|
+
Array.from({ length: n }, (_unused, index) => index).map(
|
|
1583
|
+
async (index) => await runCandidate(opts, index, generate, edit)
|
|
1584
|
+
)
|
|
1585
|
+
);
|
|
1586
|
+
const candidates = [];
|
|
1587
|
+
let firstError;
|
|
1588
|
+
for (const result of results) {
|
|
1589
|
+
if (result.isOk()) {
|
|
1590
|
+
candidates.push(result.value);
|
|
1591
|
+
} else {
|
|
1592
|
+
firstError ??= result.error;
|
|
1593
|
+
}
|
|
1594
|
+
}
|
|
1595
|
+
if (candidates.length === 0) {
|
|
1596
|
+
return err2(
|
|
1597
|
+
firstError ?? new MotifError(`all ${n} bestOfN candidates failed`, 0)
|
|
1598
|
+
);
|
|
1599
|
+
}
|
|
1600
|
+
const winner = await selectWinner(opts, candidates);
|
|
1601
|
+
if (winner.isErr()) {
|
|
1602
|
+
return err2(winner.error);
|
|
1603
|
+
}
|
|
1604
|
+
const { chosenIndex, reason } = winner.value;
|
|
1605
|
+
const best = candidates[chosenIndex];
|
|
1606
|
+
if (best === void 0) {
|
|
1607
|
+
return err2(new MotifError("bestOfN failed to resolve a winner", 0));
|
|
1608
|
+
}
|
|
1609
|
+
const totalCostUsd = Number(
|
|
1610
|
+
candidates.reduce((sum, candidate) => sum + candidate.cost.usd, 0).toFixed(6)
|
|
1611
|
+
);
|
|
1612
|
+
return ok2({
|
|
1613
|
+
best,
|
|
1614
|
+
chosenIndex,
|
|
1615
|
+
...reason === void 0 ? {} : { reason },
|
|
1616
|
+
candidates,
|
|
1617
|
+
totalCostUsd
|
|
1618
|
+
});
|
|
1619
|
+
} catch (error) {
|
|
1620
|
+
return err2(toMotifError(error));
|
|
1621
|
+
}
|
|
1538
1622
|
}
|
|
1539
1623
|
function defaultResolveModel(provider, modelId, apiKey) {
|
|
1540
1624
|
return getProviderAdapter(provider).resolveModel(modelId, apiKey);
|