Execution attempt

Sign in with GitHub
← Experiment E11 · Does Qwen2.5-1.5B score the same multiple-choice content lower in Polish than in English, and do Polish-heavy continued pretraining and the APT4 transplant change that gap?

Execution attempt · A109 · planned

Builds the translation-paired Belebele and MMLU item files from the seeded source samples and gates pairing and gold parity.

Succeeded · Registered by @stw2 via agent. Reported and received times are kept apart; no computation or result is verified.

Pinned source and configuration

https://github.com/stw2/tokenizer-science-tax @ 3c128813c117fd41b46d3a9949c2c8751134df27

Reference checked 2026-09-14 13:17 UTC. Later commits, branches or plan changes do not retarget this attempt.

Command
python scripts/01_prepare_items.py
Working directory
experiments/E11-cross-language-access
Configuration paths
None
Parameters
sample seed 42; 600 items per benchmark; hard gates on pairing, scaffold and gold parity
Environment
3.14.5; pandas and pyarrow; no model; Apple silicon, 128 GB unified memory
Output directory
Not recorded

Inputs

Materials the registrant named when registering this attempt, by content identity; obtainability is derived from their location reports. Nothing is fetched or verified.

Delivered events

  1. Registered

    #1

    Builds the translation-paired Belebele and MMLU item files from the seeded source samples and gates pairing and gold parity.

    reported · received · @stw2 via agent · posted to the Thread

  2. Started

    #2

    Started.

    reported · received · @stw2 via agent · attempt only

  3. Succeeded

    #3

    Completed with exit code 0.

    Exit code 0.

    • belebele_en.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/3c128813c117fd41b46d3a9949c2c8751134df27/experiments/E11-cross-language-access/data/belebele_en.jsonl · public · sha256 db7d6cddffa0… · 461133 bytes · M485
    • belebele_pl.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/3c128813c117fd41b46d3a9949c2c8751134df27/experiments/E11-cross-language-access/data/belebele_pl.jsonl · public · sha256 f60e5d4580f2… · 501994 bytes · M486
    • mmlu_en.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/3c128813c117fd41b46d3a9949c2c8751134df27/experiments/E11-cross-language-access/data/mmlu_en.jsonl · public · sha256 be09c7756aaf… · 357133 bytes · M487
    • mmlu_pl.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/3c128813c117fd41b46d3a9949c2c8751134df27/experiments/E11-cross-language-access/data/mmlu_pl.jsonl · public · sha256 ee3b76d0e983… · 376863 bytes · M488
    • manifest.json · Archived private repository, experiments/E11/data/manifest.json at commit ae11e89; the public copy is regenerated by the rebuild attempt · unavailable · sha256 1979ab1bad0b… · 1092 bytes · M489

    reported · received · @stw2 via agent · posted to the Thread

Report an event

The registrant’s capture tool normally delivers start and outcome events. Reporting here is the same author report with the browser as the reported time; it does not observe the process.

This attempt has a delivered outcome. A new execution is a new attempt.