Datasets:

SoroushVahidi
/

mwfas-heuristic-metrics

source_config large_stringclasses 3 values	experiment_family large_stringclasses 3 values	algorithm large_stringclasses 2 values	configuration_id large_stringlengths 5 28	split large_stringclasses 2 values	instance_family large_stringclasses 6 values	benchmark_resource large_stringclasses 1 value	n_runs int64 1 980	n_instances int64 1 49	objective_weight_mean float64 1 696k	objective_weight_median float64 1 696k	normalized_objective_mean float64 0.01 18.5k	runtime_seconds_mean float64 0 231	runtime_seconds_median float64 0 43.8	success_rate float64 1 1	validation_rate float64 1 1
run_metrics	coap_ipsns_holdout	IPSNS	iters_10	holdout	core	alidasdan/graph-benchmarks	25	5	13,317	1,688	0.398259	0.001501	0.001573	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_10	holdout	core-big	alidasdan/graph-benchmarks	10	2	673,200.5	673,200.5	0.049437	1.306975	1.305159	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_10	holdout	iscas	alidasdan/graph-benchmarks	50	10	159,138.12	62,930	0.008852	19.995601	0.594806	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_10	holdout	iscas-small	alidasdan/graph-benchmarks	40	8	20,519.375	9,818	0.008924	0.042455	0.025348	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_10	tuning	core	alidasdan/graph-benchmarks	15	3	16,976.333333	18,638	0.206677	0.001074	0.001066	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_10	tuning	iscas	alidasdan/graph-benchmarks	25	5	25,856	16,618	0.017476	0.073238	0.022654	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_10	tuning	iscas-small	alidasdan/graph-benchmarks	50	10	7,753.9	6,365	0.012179	0.007157	0.007129	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_400	holdout	core	alidasdan/graph-benchmarks	25	5	13,317	1,688	0.398259	0.020195	0.016928	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_400	holdout	core-big	alidasdan/graph-benchmarks	10	2	673,200.5	673,200.5	0.049437	26.979299	26.739421	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_400	holdout	iscas	alidasdan/graph-benchmarks	50	10	159,035.1	62,930	0.00885	231.362123	8.620184	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_400	holdout	iscas-small	alidasdan/graph-benchmarks	40	8	20,519.375	9,818	0.008924	0.599796	0.430542	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_400	tuning	core	alidasdan/graph-benchmarks	15	3	16,976.333333	18,638	0.206677	0.010376	0.010574	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_400	tuning	iscas	alidasdan/graph-benchmarks	25	5	25,856	16,618	0.017476	1.288353	0.495793	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_400	tuning	iscas-small	alidasdan/graph-benchmarks	50	10	7,753.9	6,365	0.012179	0.114659	0.098462	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50	holdout	core	alidasdan/graph-benchmarks	25	5	13,317	1,688	0.398259	0.00336	0.00327	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50	holdout	core-big	alidasdan/graph-benchmarks	10	2	673,200.5	673,200.5	0.049437	3.938488	3.937322	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50	holdout	iscas	alidasdan/graph-benchmarks	50	10	159,074.62	62,930	0.008851	41.161098	1.549165	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50	holdout	iscas-small	alidasdan/graph-benchmarks	40	8	20,519.375	9,818	0.008924	0.099394	0.066439	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50	tuning	core	alidasdan/graph-benchmarks	15	3	16,976.333333	18,638	0.206677	0.002003	0.002029	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50	tuning	iscas	alidasdan/graph-benchmarks	25	5	25,856	16,618	0.017476	0.196931	0.070886	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50	tuning	iscas-small	alidasdan/graph-benchmarks	50	10	7,753.9	6,365	0.012179	0.018195	0.016681	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_addback25	holdout	core	alidasdan/graph-benchmarks	25	5	13,317	1,688	0.398259	0.003374	0.00331	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_addback25	holdout	core-big	alidasdan/graph-benchmarks	10	2	670,209	670,209	0.049398	3.907404	3.901923	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_addback25	holdout	iscas	alidasdan/graph-benchmarks	50	10	159,131.2	63,095	0.008857	40.359317	1.503523	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_addback25	holdout	iscas-small	alidasdan/graph-benchmarks	40	8	20,503	9,818	0.0089	0.098443	0.070516	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_addback25	tuning	core	alidasdan/graph-benchmarks	15	3	16,976.333333	18,638	0.206677	0.002024	0.002046	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_addback25	tuning	iscas	alidasdan/graph-benchmarks	25	5	25,856	16,618	0.017476	0.178505	0.07059	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_addback25	tuning	iscas-small	alidasdan/graph-benchmarks	50	10	7,718.6	6,365	0.012127	0.01794	0.016376	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_topk5	holdout	core	alidasdan/graph-benchmarks	25	5	13,317	1,688	0.398259	0.00336	0.003278	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_topk5	holdout	core-big	alidasdan/graph-benchmarks	10	2	673,200.5	673,200.5	0.049437	3.934104	3.928639	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_topk5	holdout	iscas	alidasdan/graph-benchmarks	50	10	159,148.2	62,930	0.008852	41.563995	1.553922	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_topk5	holdout	iscas-small	alidasdan/graph-benchmarks	40	8	20,519.375	9,818	0.008924	0.101458	0.068768	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_topk5	tuning	core	alidasdan/graph-benchmarks	15	3	16,976.333333	18,638	0.206677	0.002019	0.002064	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_topk5	tuning	iscas	alidasdan/graph-benchmarks	25	5	25,856	16,618	0.017476	0.197586	0.071429	1	1
run_metrics	coap_ipsns_holdout	IPSNS	iters_50_topk5	tuning	iscas-small	alidasdan/graph-benchmarks	50	10	7,753.9	6,365	0.012179	0.018317	0.01673	1	1
run_metrics	coap_ipsns_holdout	IPSNS	seed_only	holdout	core	alidasdan/graph-benchmarks	25	5	13,317	1,688	0.398259	0.000902	0.0009	1	1
run_metrics	coap_ipsns_holdout	IPSNS	seed_only	holdout	core-big	alidasdan/graph-benchmarks	10	2	696,473	696,473	0.051161	0.634568	0.636235	1	1
run_metrics	coap_ipsns_holdout	IPSNS	seed_only	holdout	iscas	alidasdan/graph-benchmarks	50	10	159,831.2	64,092	0.008928	14.459304	0.351566	1	1
run_metrics	coap_ipsns_holdout	IPSNS	seed_only	holdout	iscas-small	alidasdan/graph-benchmarks	40	8	20,587.5	9,891	0.009009	0.028277	0.016201	1	1
run_metrics	coap_ipsns_holdout	IPSNS	seed_only	tuning	core	alidasdan/graph-benchmarks	15	3	16,976.333333	18,638	0.206677	0.000795	0.000786	1	1
run_metrics	coap_ipsns_holdout	IPSNS	seed_only	tuning	iscas	alidasdan/graph-benchmarks	25	5	25,856	16,618	0.017476	0.042144	0.010086	1	1
run_metrics	coap_ipsns_holdout	IPSNS	seed_only	tuning	iscas-small	alidasdan/graph-benchmarks	50	10	7,753.9	6,365	0.012179	0.004432	0.004523	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	addback_0.15	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.231834	0.231834	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	addback_0.15	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.009463	0.009312	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	addback_0.15	null	core-big	alidasdan/graph-benchmarks	2	2	18,489	18,489	18,489	23.562261	23.562261	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	addback_0.25	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.240791	0.240791	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	addback_0.25	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.009419	0.009238	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	addback_0.25	null	core-big	alidasdan/graph-benchmarks	2	2	18,468	18,468	18,468	26.586927	26.586927	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	addback_0.35	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.234439	0.234439	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	addback_0.35	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.009738	0.009064	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	addback_0.35	null	core-big	alidasdan/graph-benchmarks	2	2	18,531	18,531	18,531	29.695048	29.695048	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	baseline_default	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.240588	0.240588	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	baseline_default	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.039368	0.009213	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	baseline_default	null	core-big	alidasdan/graph-benchmarks	2	2	18,504.5	18,504.5	18,504.5	29.0403	29.0403	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	iters_100	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.001484	0.001484	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	iters_100	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.002946	0.002868	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	iters_100	null	core-big	alidasdan/graph-benchmarks	2	2	18,504.5	18,504.5	18,504.5	7.56169	7.56169	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	iters_200	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.002653	0.002653	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	iters_200	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.005089	0.004823	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	iters_200	null	core-big	alidasdan/graph-benchmarks	2	2	18,504.5	18,504.5	18,504.5	14.746764	14.746764	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	iters_600	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.00573	0.00573	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	iters_600	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.013703	0.013277	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	iters_600	null	core-big	alidasdan/graph-benchmarks	2	2	18,504.5	18,504.5	18,504.5	43.830538	43.830538	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	remove_0.01	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.004655	0.004655	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	remove_0.01	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.00943	0.009015	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	remove_0.01	null	core-big	alidasdan/graph-benchmarks	2	2	18,490.5	18,490.5	18,490.5	28.747062	28.747062	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	remove_0.05	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.00457	0.00457	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	remove_0.05	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.009523	0.009038	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	remove_0.05	null	core-big	alidasdan/graph-benchmarks	2	2	18,506.5	18,506.5	18,506.5	28.254481	28.254481	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	rng_2	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.004574	0.004574	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	rng_2	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.009355	0.008868	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	rng_2	null	core-big	alidasdan/graph-benchmarks	2	2	18,504.5	18,504.5	18,504.5	29.315242	29.315242	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	rng_3	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.003939	0.003939	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	rng_3	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.00947	0.009306	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	rng_3	null	core-big	alidasdan/graph-benchmarks	2	2	18,504.5	18,504.5	18,504.5	29.297106	29.297106	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	topk_10	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.003908	0.003908	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	topk_10	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.009371	0.009076	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	topk_10	null	core-big	alidasdan/graph-benchmarks	2	2	18,504.5	18,504.5	18,504.5	29.193273	29.193273	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	topk_20	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.003839	0.003839	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	topk_20	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.009356	0.009002	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	topk_20	null	core-big	alidasdan/graph-benchmarks	2	2	18,504.5	18,504.5	18,504.5	28.941193	28.941193	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	topk_5	null	core	alidasdan/graph-benchmarks	1	1	1	1	1	0.003823	0.003823	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	topk_5	null	core-bad	alidasdan/graph-benchmarks	7	7	768.857143	770	768.857143	0.009434	0.009182	1	1
sensitivity	coap_ipsns_sensitivity	IPSNS	topk_5	null	core-big	alidasdan/graph-benchmarks	2	2	18,504.5	18,504.5	18,504.5	28.657234	28.657234	1	1
robustness	exp10_stochastic_robustness	DRMacIver/FAS	drmaciver_robustness_20_reps	null	core	alidasdan/graph-benchmarks	980	49	6,291.977551	1,698	0.294237	0.031462	0.004816	1	1
robustness	exp10_stochastic_robustness	DRMacIver/FAS	drmaciver_robustness_20_reps	null	core-bad	alidasdan/graph-benchmarks	140	7	816.4	770	0.050999	0.028038	0.007709	1	1
robustness	exp10_stochastic_robustness	DRMacIver/FAS	drmaciver_robustness_20_reps	null	core-big	alidasdan/graph-benchmarks	80	4	593,795.625	386,638.5	0.078964	6.132015	6.128918	1	1
robustness	exp10_stochastic_robustness	DRMacIver/FAS	drmaciver_robustness_20_reps	null	iscas	alidasdan/graph-benchmarks	260	13	125,181.111538	104,567.5	0.028006	20.897301	12.142217	1	1
robustness	exp10_stochastic_robustness	DRMacIver/FAS	drmaciver_robustness_20_reps	null	iscas-small	alidasdan/graph-benchmarks	360	18	38,323.077778	23,954.5	0.027311	3.923591	2.188667	1	1
robustness	exp10_stochastic_robustness	DRMacIver/FAS	drmaciver_robustness_20_reps	null	tests	alidasdan/graph-benchmarks	40	2	20	20	0.058824	0.004748	0.004785	1	1
robustness	exp10_stochastic_robustness	IPSNS	ipsns_robustness_20_reps	null	core	alidasdan/graph-benchmarks	980	49	6,255.122449	1,688	0.294039	0.009587	0.006013	1	1
robustness	exp10_stochastic_robustness	IPSNS	ipsns_robustness_20_reps	null	core-bad	alidasdan/graph-benchmarks	140	7	768.857143	770	0.049612	0.008932	0.008567	1	1
robustness	exp10_stochastic_robustness	IPSNS	ipsns_robustness_20_reps	null	core-big	alidasdan/graph-benchmarks	80	4	345,852.5	159,152.5	0.050768	27.223803	26.064149	1	1
robustness	exp10_stochastic_robustness	IPSNS	ipsns_robustness_20_reps	null	iscas	alidasdan/graph-benchmarks	260	13	62,653.384615	40,635	0.012192	5.188082	2.350815	1	1
robustness	exp10_stochastic_robustness	IPSNS	ipsns_robustness_20_reps	null	iscas-small	alidasdan/graph-benchmarks	360	18	13,427.444444	9,044.5	0.010732	0.322738	0.10975	1	1
robustness	exp10_stochastic_robustness	IPSNS	ipsns_robustness_20_reps	null	tests	alidasdan/graph-benchmarks	40	2	20	20	0.058824	0.003056	0.003175	1	1

MWFAS Heuristic Metrics

This is a metrics-only dataset for experiments on the Minimum Weighted Feedback Arc Set (MWFAS) problem and its connection to ranking from pairwise comparisons.

The dataset contains project-generated run-level outcomes for MWFAS heuristic experiments. It does not include raw third-party graph benchmark files, graph edge lists, adjacency lists, raw ranking vectors, manuscript drafts, logs, or machine-local execution artifacts.

Canonical Hugging Face repository: SoroushVahidi/mwfas-heuristic-metrics. MWFAS Heuristic Metrics v1 is published and available here.

What is in this dataset?

A row represents one sanitized experimental outcome from one of three canonical experiment families:

run_metrics: one IPSNS parameter-grid run on a graph instance from the COAP holdout/tuning study.
sensitivity: one one-at-a-time IPSNS parameter-sensitivity run.
robustness: one repeated stochastic robustness run for IPSNS or DRMacIver/FAS on the common sparse benchmark subset.
aggregate_summary: reproducible aggregates derived from the three public primary configs.

The main scientific question is: how do MWFAS heuristics behave across graph instances, parameter configurations, and repeated runs in terms of objective quality, runtime, improvement over incumbents, and validation status?

Problem background

A feedback arc set is a set of directed arcs whose removal makes a directed graph acyclic. In the weighted version, each arc has a weight and the objective is to minimize the total weight of arcs that point backward under the produced ordering. Lower objective values are better.

The problem is also connected to ranking from pairwise comparisons: a weighted directed edge can represent one item being preferred over another, and minimizing backward weight corresponds to finding an ordering with small weighted disagreement.

Configurations

`run_metrics`

Rows: 1,290.

Source: experiments/coap_ipsns_holdout/results/runs.jsonl from the source repository.

This config contains the Stage-2 IPSNS tuning/holdout validation grid. Rows include sanitized instance identifiers, graph structural metadata, algorithm configuration, seed, objective, runtime, improvement over the best initial seed, and validation status.

`sensitivity`

Rows: 140.

Source: experiments/coap_ipsns_sensitivity/summary/canonical_runs.csv.

This config contains Stage-1 one-at-a-time parameter-sensitivity runs. Rows identify the varied parameter, configuration, objective, runtime, and improvement relative to the default/baseline settings.

`robustness`

Rows: 3,720.

Source: experiments/exp10_stochastic_robustness/summary/run_level_results.csv.

This config contains repeated-run stochastic robustness data: 1,860 IPSNS runs and 1,860 DRMacIver/FAS runs over a common sparse subset. It supports analysis of run-to-run stability, runtime, objective quality, and validation outcomes.

`aggregate_summary`

Rows: 96.

This config is derived only from the public primary configs. It groups by experiment family, algorithm, configuration, split where applicable, benchmark resource, and instance family, then reports run counts, instance counts, objective summaries, runtime summaries, success rate, and validation rate.

Important fields

benchmark_resource: upstream benchmark collection referenced by the row. Raw benchmark files are not redistributed here.
instance_family: sanitized family/group inferred from the source benchmark path, such as core, core-bad, or iscas.
instance_id: public benchmark-local instance identifier. It is not an edge list or graph encoding.
n_vertices, n_edges: graph structural metadata.
density: directed density computed as n_edges / (n_vertices * (n_vertices - 1)) when applicable.
algorithm: algorithm or baseline family, e.g. IPSNS or DRMacIver/FAS.
configuration_id: public configuration label.
objective_weight: weighted backward-arc objective. Lower is better.
normalized_objective: objective divided by total edge weight when available. Lower is better.
total_edge_weight: total weight of graph arcs used by the experiment.
runtime_seconds: measured experiment runtime in seconds. Interpret as implementation/runtime-environment dependent, not a hardware-independent complexity measure.
initial_incumbent_objective: best initial objective available before refinement, usually the best of LR-TA and WMSF seeds where recorded.
improvement_absolute: initial_incumbent_objective - objective_weight where available. Positive means the run improved the incumbent.
improvement_relative: improvement divided by the initial incumbent objective where available.
validated: boolean validation flag derived from source status and, for robustness rows, ordering/objective/acyclicity checks.
source_commit: source repository commit associated with the record where recoverable.

What is excluded?

This release intentionally excludes:

raw benchmark graph files;
graph edge lists and adjacency matrices;
temporary ranking vectors;
raw output paths and local filenames;
machine names, CPU/environment details, PIDs, and process metadata;
manuscript drafts, reviewer/editorial material, logs, and private build-audit files;
the separate Ranking-by-FAS/GNNRank result matrices from the separate local Ranking-by-FAS project.

Benchmark provenance

The included rows reference graph benchmark instances primarily from the alidasdan/graph-benchmarks resource. This dataset redistributes only project-generated metrics and derived structural metadata; it does not redistribute the graph benchmark files themselves.

Users who need the underlying graph instances should obtain them from the canonical upstream source and follow its terms and citation guidance.

Intended uses

Appropriate uses include:

reproducing and auditing MWFAS heuristic experiment summaries;
analyzing runtime/objective tradeoffs across graph instances and configurations;
studying parameter sensitivity and stochastic robustness;
meta-analysis or meta-learning over algorithm outcomes using structural graph metadata;
comparing heuristic stability and validation outcomes without rerunning all experiments.

Non-intended uses

This dataset should not be used as:

a replacement for the upstream graph benchmark datasets;
a source of graph edge lists or raw ranking data;
a universal benchmark of all FAS/MWFAS algorithms;
a hardware-independent runtime leaderboard;
evidence that one method dominates outside the stated benchmark families and protocols.

Limitations

Runtime values depend on implementation and execution environment.
The released metrics are historical results from specific code/protocol versions.
The run_metrics and sensitivity configs emphasize IPSNS parameter behavior, not a full cross-algorithm benchmark.
The robustness config is quality-focused and not an equal-time comparison.
Raw graph inputs are excluded; users must fetch upstream graph benchmarks separately to rerun algorithms from scratch.

How This Dataset Differs from Existing Resources

Raw graph benchmark collections such as graph-benchmarks, LOLIB, and SNAP Wiki-Vote provide graph instances or network data. GNNRank-style resources provide ranking datasets, code, or paper-level result matrices. This release provides reusable run-level MWFAS heuristic metrics, parameter-sensitivity records, and robustness measurements derived from experiments, without redistributing raw graph inputs.

The searched public resources primarily provide graph instances, code, or paper-level summaries; this release provides a compact tabular representation of project-generated algorithm-run outcomes.

Relation to Soroush Vahidi's existing Hugging Face datasets

This is a graph-optimization / weighted feedback-arc-set experimental metrics dataset. It is distinct from:

SoroushVahidi/lafc-evict, which contains cache-eviction candidate supervision;
SoroushVahidi/module-intervention-credit, which contains LLM-serving scheduler intervention data;
SoroushVahidi/consistency-aware-judgments, which contains IR pairwise LLM judgment data;
SoroushVahidi/frontier-allocation-metrics, which contains budgeted LLM inference cost/latency/correctness outcomes;
SoroushVahidi/scidocs, which is a third-party BEIR/SciDocs mirror;
SoroushVahidi/lafc-evict-sample, which is a synthetic workflow artifact.

Associated paper

Primary associated paper:

Soroush Vahidi and Ioannis Koutis. “Minimum Weighted Feedback Arc Sets for Ranking from Pairwise Comparisons.” arXiv:2412.16181. DOI: 10.48550/arXiv.2412.16181.

The paper introduces and studies the MWFAS/ranking methodology and reports algorithmic results. Cite the dataset when using the released metrics. Cite the paper when discussing the methodology or scientific findings. Cite both when using the data and the associated methodology/results.

Related supporting work:

Soroush Vahidi. “Incumbent-Protected SCC-Neighborhood Search for the Weighted Feedback Arc Set Problem.” SSRN abstract 6281222. This is treated as supporting related work for the IPSNS/SCC-neighborhood component, not as the primary dataset citation.

License

CC BY 4.0 for the released project-generated metrics, summaries, metadata, and documentation to the extent controlled by this project.

This license does not relicense upstream graph benchmark datasets, third-party software, or papers referenced by identifier or citation.

Version and immutable revision

Version: v1.

Immutable scientific release revision: 5621684d3c03138d5b2ebe544e91ae5698e67b5a.

For exact reproducibility, cite or record the Hugging Face repository together with this revision. The Parquet data files for v1 were first published at this revision.

Recommended citation

Vahidi, S. (2026). MWFAS Heuristic Metrics: Run-Level Outcomes for Minimum Weighted Feedback Arc Set Experiments (v1) [Data set]. Hugging Face. https://huggingface.co/datasets/SoroushVahidi/mwfas-heuristic-metrics

No dataset DOI has been assigned. For exact reproducibility, cite the immutable Hugging Face revision 5621684d3c03138d5b2ebe544e91ae5698e67b5a.

Downloads last month: 39

Paper for SoroushVahidi/mwfas-heuristic-metrics

Minimum Weighted Feedback Arc Sets for Ranking from Pairwise Comparisons

Paper • 2412.16181 • Published Dec 4, 2025