@tangle-network/agent-eval 0.110.0 → 0.111.0
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/CHANGELOG.md +18 -0
- package/dist/belief-state/index.js +1 -1
- package/dist/benchmarks/index.js +6 -6
- package/dist/builder-eval/index.js +1 -1
- package/dist/campaign/index.d.ts +50 -5
- package/dist/campaign/index.js +7 -5
- package/dist/{chunk-X6NIXVOD.js → chunk-3LDPUYAC.js} +2 -2
- package/dist/{chunk-N22ZO7FV.js → chunk-7QPK6W7R.js} +2 -2
- package/dist/{chunk-QFGTU7MT.js → chunk-BOLYPSAE.js} +4 -4
- package/dist/{chunk-RNB2NICW.js → chunk-D2XDWRFX.js} +74 -4
- package/dist/chunk-D2XDWRFX.js.map +1 -0
- package/dist/{chunk-6MFFKPZ4.js → chunk-EMKORATZ.js} +2 -2
- package/dist/{chunk-GTERJI6Q.js → chunk-RCXMWGRY.js} +16 -5
- package/dist/{chunk-GTERJI6Q.js.map → chunk-RCXMWGRY.js.map} +1 -1
- package/dist/{chunk-6HFYGEZ3.js → chunk-RT2AFUXM.js} +2 -2
- package/dist/{chunk-XMSYF4A7.js → chunk-VC43KQCK.js} +24 -1
- package/dist/chunk-VC43KQCK.js.map +1 -0
- package/dist/{chunk-T6W5ADLG.js → chunk-VLNGPUJJ.js} +2 -2
- package/dist/{chunk-6PL5MGDL.js → chunk-WZOIFFG2.js} +2 -2
- package/dist/{chunk-3PFZBGMR.js → chunk-XNCF3JU3.js} +2 -2
- package/dist/{chunk-PLOMR3HP.js → chunk-YWGKQARF.js} +2 -2
- package/dist/contract/index.d.ts +11 -10
- package/dist/contract/index.js +6 -6
- package/dist/contract/index.js.map +1 -1
- package/dist/{gepa-BUNP3606.d.ts → gepa-T8T215nw.d.ts} +6 -4
- package/dist/index.d.ts +122 -5
- package/dist/index.js +470 -210
- package/dist/index.js.map +1 -1
- package/dist/meta-eval/index.d.ts +1 -1
- package/dist/meta-eval/index.js +2 -2
- package/dist/openapi.json +1 -1
- package/dist/pipelines/index.js +2 -2
- package/dist/{provenance-DMvsfknv.d.ts → provenance-KhY8ESVM.d.ts} +2 -2
- package/dist/reporting.d.ts +1 -1
- package/dist/reporting.js +4 -4
- package/dist/rl.js +5 -5
- package/dist/{run-campaign-PNHAJIT2.js → run-campaign-HJQEDCJQ.js} +3 -3
- package/dist/{statistics-D88peojY.d.ts → statistics-CDfpwIdX.d.ts} +13 -1
- package/docs/improvement-glossary.md +14 -13
- package/package.json +1 -1
- package/dist/chunk-RNB2NICW.js.map +0 -1
- package/dist/chunk-XMSYF4A7.js.map +0 -1
- /package/dist/{chunk-X6NIXVOD.js.map → chunk-3LDPUYAC.js.map} +0 -0
- /package/dist/{chunk-N22ZO7FV.js.map → chunk-7QPK6W7R.js.map} +0 -0
- /package/dist/{chunk-QFGTU7MT.js.map → chunk-BOLYPSAE.js.map} +0 -0
- /package/dist/{chunk-6MFFKPZ4.js.map → chunk-EMKORATZ.js.map} +0 -0
- /package/dist/{chunk-6HFYGEZ3.js.map → chunk-RT2AFUXM.js.map} +0 -0
- /package/dist/{chunk-T6W5ADLG.js.map → chunk-VLNGPUJJ.js.map} +0 -0
- /package/dist/{chunk-6PL5MGDL.js.map → chunk-WZOIFFG2.js.map} +0 -0
- /package/dist/{chunk-3PFZBGMR.js.map → chunk-XNCF3JU3.js.map} +0 -0
- /package/dist/{chunk-PLOMR3HP.js.map → chunk-YWGKQARF.js.map} +0 -0
- /package/dist/{run-campaign-PNHAJIT2.js.map → run-campaign-HJQEDCJQ.js.map} +0 -0
|
@@ -5,7 +5,7 @@ import {
|
|
|
5
5
|
pairedBootstrap,
|
|
6
6
|
pairedMde,
|
|
7
7
|
wilcoxonSignedRank
|
|
8
|
-
} from "./chunk-
|
|
8
|
+
} from "./chunk-VC43KQCK.js";
|
|
9
9
|
import {
|
|
10
10
|
canonicalize,
|
|
11
11
|
hashJson
|
|
@@ -877,4 +877,4 @@ export {
|
|
|
877
877
|
RESEARCH_REPORT_HARD_PAIR_FLOOR,
|
|
878
878
|
researchReport
|
|
879
879
|
};
|
|
880
|
-
//# sourceMappingURL=chunk-
|
|
880
|
+
//# sourceMappingURL=chunk-EMKORATZ.js.map
|
|
@@ -1,13 +1,13 @@
|
|
|
1
1
|
import {
|
|
2
2
|
runCampaign,
|
|
3
3
|
summarizeBackendIntegrity
|
|
4
|
-
} from "./chunk-
|
|
4
|
+
} from "./chunk-XNCF3JU3.js";
|
|
5
5
|
import {
|
|
6
6
|
detectRewardHacking
|
|
7
|
-
} from "./chunk-
|
|
7
|
+
} from "./chunk-7QPK6W7R.js";
|
|
8
8
|
import {
|
|
9
9
|
pairedBootstrap
|
|
10
|
-
} from "./chunk-
|
|
10
|
+
} from "./chunk-VC43KQCK.js";
|
|
11
11
|
import {
|
|
12
12
|
DEFAULT_REDACTION_RULES
|
|
13
13
|
} from "./chunk-GGE4NNQT.js";
|
|
@@ -1953,6 +1953,17 @@ async function runOptimization(opts) {
|
|
|
1953
1953
|
const scored = [
|
|
1954
1954
|
toParetoParent(opts.baselineSurface, winnerSurfaceHash, baselineCampaign, -1)
|
|
1955
1955
|
];
|
|
1956
|
+
if (opts.analyzeGeneration && opts.maxGenerations > 0 && baselineCampaign.cells.length > 0) {
|
|
1957
|
+
const fresh = await opts.analyzeGeneration({
|
|
1958
|
+
generation: -1,
|
|
1959
|
+
runDir: `${opts.runDir}/baseline`,
|
|
1960
|
+
candidates: [
|
|
1961
|
+
{ surfaceHash: winnerSurfaceHash, campaign: baselineCampaign, composite: winnerComposite }
|
|
1962
|
+
],
|
|
1963
|
+
history
|
|
1964
|
+
});
|
|
1965
|
+
if (Array.isArray(fresh)) currentFindings = fresh;
|
|
1966
|
+
}
|
|
1956
1967
|
for (let gen = 0; gen < opts.maxGenerations; gen++) {
|
|
1957
1968
|
if (proposer.decide?.({ history }).stop) break;
|
|
1958
1969
|
const paretoParents = computeParetoFrontier(scored);
|
|
@@ -2124,7 +2135,7 @@ async function runImprovementLoop(opts) {
|
|
|
2124
2135
|
const dispatchTimeoutMs = opts.dispatchTimeoutMs ?? DEFAULT_DISPATCH_TIMEOUT_MS;
|
|
2125
2136
|
const optimization = await runOptimization({ ...opts, dispatchTimeoutMs });
|
|
2126
2137
|
const winnerIsBaseline = optimization.winnerSurfaceHash === surfaceHash(opts.baselineSurface);
|
|
2127
|
-
const { runCampaign: runCampaign2 } = await import("./run-campaign-
|
|
2138
|
+
const { runCampaign: runCampaign2 } = await import("./run-campaign-HJQEDCJQ.js");
|
|
2128
2139
|
const baselineOnHoldout = await runCampaign2({
|
|
2129
2140
|
...opts,
|
|
2130
2141
|
dispatchTimeoutMs,
|
|
@@ -2578,4 +2589,4 @@ export {
|
|
|
2578
2589
|
provenanceSpansPath,
|
|
2579
2590
|
emitLoopProvenance
|
|
2580
2591
|
};
|
|
2581
|
-
//# sourceMappingURL=chunk-
|
|
2592
|
+
//# sourceMappingURL=chunk-RCXMWGRY.js.map
|