trace-bench v0.3.0

A simulated microservice trace benchmark with ground-truth causal graphs. Each named instance ships a raw heterogeneous log feed (no parent pointers), the correlated sequence views, the mechanism graph that generated the data (latent state variables flagged), its latent projection over (operation, outcome) tokens as the scoring target, the deployment call topology as a candidate structural prior, injected-fault labels and the oracle linkage that makes correlation loss a measured quantity. Every corpus is wholly synthetic and reproduces byte-identically from its configuration, seed, tool version and constants version (see each manifest.json).

Generator (MIT): https://github.com/alex-chadyuk/trace-bench — tool version 0.3.0. Corpus licence: CC BY 4.0. Fitted realism constants: 1.

instance	variant	seed	alphabet (realized, train)	mechanism nodes	directed / bidirected at floor	edges on a Monte-Carlo estimate (max SE)
xs	latent	0	78	128	453 / 111	0 / 0 (0.0)
xs	latent	1	77	128	479 / 107	0 / 0 (0.0)
xs	latent	2	76	128	410 / 120	0 / 0 (0.0)
xs	latent	3	79	128	241 / 185	0 / 0 (0.0)
xs	latent	4	79	128	434 / 165	0 / 0 (0.0)
xs	twin	0	78	128	720 / 0	0 / 0 (0.0)
xs	twin	1	77	128	743 / 0	0 / 0 (0.0)
xs	twin	2	76	128	664 / 0	0 / 0 (0.0)
xs	twin	3	79	128	506 / 0	0 / 0 (0.0)
xs	twin	4	79	128	716 / 0	0 / 0 (0.0)
s	latent	0	322	430	2064 / 796	0 / 0 (0.0)
s	latent	1	322	430	2545 / 859	0 / 0 (0.0)
s	latent	2	322	430	2555 / 851	0 / 0 (0.0)
s	latent	3	322	430	2517 / 791	0 / 0 (0.0)
s	latent	4	322	430	2576 / 997	0 / 0 (0.0)
s	twin	0	322	430	3114 / 0	0 / 0 (0.0)
s	twin	1	322	430	3655 / 0	0 / 0 (0.0)
s	twin	2	322	430	3646 / 0	0 / 0 (0.0)
s	twin	3	322	430	3559 / 0	0 / 0 (0.0)
s	twin	4	322	430	3649 / 0	0 / 0 (0.0)
m	latent	0	2644	2942	24639 / 11878	0 / 6879 (0.002495237)
m	latent	1	2650	2942	24205 / 14597	0 / 7149 (0.001746237)
m	latent	2	2654	2942	25758 / 13387	0 / 8588 (0.002343441)
m	latent	3	2650	2942	24527 / 13172	0 / 7231 (0.001926874)
m	latent	4	2653	2942	26436 / 14993	0 / 27761 (0.002130168)
m	twin	0	2644	2942	33265 / 0	0 / 0 (0.0)
m	twin	1	2650	2942	32952 / 0	0 / 0 (0.0)
m	twin	2	2654	2942	34420 / 0	6 / 0 (0.0)
m	twin	3	2650	2942	33207 / 0	0 / 0 (0.0)
m	twin	4	2653	2942	35209 / 0	0 / 0 (0.0)
l	latent	0	7435	8382	73837 / 50291	0 / 23984 (0.002499367)
l	latent	1	7513	8382	73107 / 58068	0 / 23703 (0.002497846)
l	latent	2	7533	8382	72764 / 60578	0 / 24375 (0.002487043)
l	latent	3	7477	8382	68885 / 56822	0 / 24311 (0.002447562)
l	latent	4	7430	8382	71266 / 58899	0 / 24374 (0.002498424)
l	twin	0	7435	8382	98678 / 0	6 / 0 (0.002499133)
l	twin	1	7513	8382	98141 / 0	6 / 0 (0.002497846)
l	twin	2	7533	8382	97944 / 0	6 / 0 (0.002487043)
l	twin	3	7477	8382	93963 / 0	6 / 0 (0.002447562)
l	twin	4	7430	8382	96285 / 0	6 / 0 (0.002476752)
xl	latent	0	19677	22772	198148 / 181365	0 / 63115 (0.002396905)
xl	latent	1	19772	22772	197410 / 201988	0 / 64834 (0.002407846)
xl	latent	2	19873	22772	193085 / 204374	0 / 64891 (0.001854232)
xl	latent	3	19788	22772	193375 / 203924	0 / 63779 (0.002499959)
xl	latent	4	19713	22772	192821 / 208923	0 / 64139 (0.002360767)
xl	twin	0	19677	22772	267210 / 0	6 / 0 (0.002396905)
xl	twin	1	19772	22772	266771 / 0	6 / 0 (0.002407251)
xl	twin	2	19873	22772	262139 / 0	6 / 0 (0.001852761)
xl	twin	3	19788	22772	262667 / 0	6 / 0 (0.002488808)
xl	twin	4	19713	22772	261924 / 0	6 / 0 (0.002360767)

To score a method: read only raw/ and views/; score against graphs/scoring-target.json (request grain) with python -m tracebench.score.

Scoring-target strengths are exact marginalisations of the mechanism, except where the exact frontier of one (source, destination) chain exceeds the tool's particle cap: those effects are counter-keyed Monte-Carlo estimates with common random numbers (the sample count and the largest standard error are recorded per edge as mc: {n, se} and per target as n_effects_mc / mc_se_max). The last column counts the request-grain edges that carry the flag; a reader who wants only exact strengths can drop them. The scorer reads strengths only.

The dataset viewer is disabled: a corpus is a tree of gzipped JSON lines and parquet files with several schemas, not one table.