Staged Factorial Screening for Budget-Constrained Micro-Pretraining

[2606.05186] Staged Factorial Screening for Budget-Constrained Micro-Pretraining

[Submitted on 27 Apr 2026]

Title:Staged Factorial Screening for Budget-Constrained Micro-Pretraining

View a PDF of the paper titled Staged Factorial Screening for Budget-Constrained Micro-Pretraining, by Felipe Chavarro Polania

View PDF HTML (experimental)

Abstract:Budget-constrained micro-pretraining often requires triaging many candidate recipes on a shared accelerator before larger search budgets are spent. We study whether a staged fractional-factorial workflow can recover stable early effect structure in this setting. On a fixed autoresearch-derived single-GPU training loop, we run 613 experiments across pilot and follow-up screens at 2, 5, and 10 minutes; full 16-condition seeded reruns at 5 and 10 minutes; targeted seeded anchor checks; same-host greedy and matched-cost random baselines; a 60-minute bridge package; and bounded Windows A100 and Linux L40S anchor continuations through 24 hours. Main penalties from total batch, depth, and width are largest at short budgets and relax as budget increases. Within the predeclared seeded full-screen families, D, A, B, and C retain non-zero estimates at 5 and 10 minutes after within-budget Benjamini-Hochberg correction, while E does not. Random search can reach strong incumbents in this 32-condition space, but repeatedly in the same low-penalty region and without factor attribution. The 60-minute bridge anchor has the lowest mean, although that package does not separate workflow refinement from the larger bridge model's capacity advantage. In bounded 12-hour and 24-hour three-anchor continuations on both hosts, the bridge has the lowest sample mean while the non-bridge ordering stays host-sensitive. We therefore present a bounded methods result: use short designed screens to identify high-penalty directions, confirm promising anchors under repeated runs, and refine locally inside the reduced space. The evidence supports a bridge-centered recommendation through 24 hours on two hosts, not hardware-invariant ranking or general hyperparameter-optimization superiority.

Comments: 23 pages, 4 figures

Subjects:

Machine Learning (cs.LG); Computation and Language (cs.CL)

Cite as: arXiv:2606.05186 [cs.LG]

(or arXiv:2606.05186v1 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2606.05186

arXiv-issued DOI via DataCite

Submission history

From: Felipe Chavarro Polania [view email] [v1] Mon, 27 Apr 2026 21:05:26 UTC (107 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Staged Factorial Screening for Budget-Constrained Micro-Pretraining, by Felipe Chavarro Polania

View PDF

HTML (experimental)

TeX Source

view license

Ancillary-file links:

Ancillary files (details):

README.txt

analysis/bridge_60min_analysis_2026-04-11.json

analysis/bridge_60min_analysis_2026-04-11_condition_summary.csv

analysis/bridge_60min_analysis_2026-04-11_pairwise_summary.csv

analysis/center_points_bridge_summary_2026-04-27.json

analysis/center_points_bridge_summary_2026-04-27_budget_summary.csv

analysis/crosshost_linux_analysis_2026-04-15.json

analysis/crosshost_linux_analysis_2026-04-15_condition_summary.csv

analysis/crosshost_linux_analysis_2026-04-15_pairwise_summary.csv

analysis/followup_abce_dlow_replicated_2026-04-09.json

analysis/followup_abce_dlow_replicated_effects_2026-04-09.csv

analysis/followup_abce_dlow_screening.json

analysis/greedy_proxy_summary.json

analysis/greedy_proxy_unique_configs.csv

analysis/linux_long_horizon_12h_analysis_2026-04-18.json

analysis/linux_long_horizon_12h_analysis_2026-04-18_condition_summary.csv

analysis/linux_long_horizon_12h_analysis_2026-04-18_pairwise_summary.csv

analysis/linux_long_horizon_24h_analysis_2026-04-24.json

analysis/linux_long_horizon_24h_analysis_2026-04-24_condition_summary.csv

analysis/linux_long_horizon_24h_analysis_2026-04-24_pairwise_summary.csv

analysis/pilot16_v1_screening.json

analysis/random_search_analysis_2026-04-14.json

analysis/random_search_analysis_2026-04-14_batch_summary.csv

analysis/random_search_analysis_2026-04-14_cumulative_summary.csv

analysis/replication_v1_bucket_summary.csv

analysis/replication_v1_source_summary.csv

analysis/replication_v1_summary.json

analysis/seed_confirmation_analysis_2026-04-10.json

analysis/seed_confirmation_analysis_2026-04-10_bucket_summary.csv

analysis/seed_confirmation_analysis_2026-04-10_condition_summary.csv

analysis/seed_confirmation_analysis_2026-04-10_dominance.csv

analysis/seeded_fullscreen_analysis_2026-04-13.json

analysis/seeded_fullscreen_analysis_2026-04-13_condition_summary.csv

analysis/seeded_fullscreen_analysis_2026-04-13_main_effect_summary.csv

analysis/windows_long_horizon_12h_analysis_2026-04-17.json

analysis/windows_long_horizon_12h_analysis_2026-04-17_condition_summary.csv

analysis/windows_long_horizon_12h_analysis_2026-04-17_pairwise_summary.csv

analysis/windows_long_horizon_24h_analysis_2026-04-24.json

analysis/windows_long_horizon_24h_analysis_2026-04-24_condition_summary.csv

analysis/windows_long_horizon_24h_analysis_2026-04-24_pairwise_summary.csv

autoresearch_seed_snapshot_manifest_2026-04-27.json

bridge_60min_base_conditions_4row.csv

center_points_bridge_9run_matrix.csv

control/README.txt

control/control_path_manifest_2026-04-27.json

control/materialize_condition.py

control/parse_train_summary.py

control/patch_prepare_time_budget.py

control/remote-env-linux.ps1

control/remote-env.ps1

control/train_windows_fallback.py

crosshost_linux_anchor_base_conditions_4row.csv

followup_abce_dlow_16run_matrix.csv

followup_abce_dlow_16run_matrix_10min.csv

full_screen_seeded_base_conditions_16row.csv

long_horizon_anchor_12h_6run.csv

long_horizon_anchor_24h_6run.csv

long_horizon_anchor_24h_base_conditions_3row.csv

long_horizon_anchor_base_conditions_3row.csv

pilot_16run_design_matrix.csv

prepare.py

pyproject.toml

replication_12run_matrix.csv

seed_confirmation_base_conditions_6row.csv

table1_run_manifest_2026-04-27.csv

table1_run_manifest_2026-04-27.json

train.py

(62 additional files not shown)

Current browse context:

cs.LG

new | recent | 2026-06

Change to browse by:

cs cs.CL

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

IArxiv recommender toggle

IArxiv Recommender (What is IArxiv?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)