Execution attempt

Sign in with GitHub
← Experiment E2 · Does APT4's tokenization of digits and number separators differ from the Mistral-derived tokenizer's, and does the difference account for the lower GSM8K score of Bielik-PL-11B-v3.0-Instruct and for accuracy differences between English and Polish number formats?

Execution attempt · A11 · planned

Converts both downloaded models to MLX bf16 without quantisation and hashes the converted files.

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 @ c43f19850803969bc1fefe537caf14e54cfd9a9f

Reference checked 2026-09-14 13:08 UTC. Later commits, branches or plan changes do not retarget this attempt.

Command
python scripts/02_convert_mlx.py
Working directory
experiments/E02-digit-policy
Configuration paths
None
Parameters
dtype bfloat16; no quantisation; mlx-lm 0.31.3 for both models; tokenizer.json re-serialised by the conversion
Environment
CPython 3.12.13 (uv 0.11.16); MLX 0.31.2 on Metal, bf16 weights, greedy decoding (temperature 0); 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

    Converts both downloaded models to MLX bf16 without quantisation and hashes the converted files.

    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.

    • bielik-pl-11b-v3.0-instruct mlx-bf16 · Not redistributed: ask the Room owner, or rebuild with scripts/02_convert_mlx.py at the pinned revisions · restricted · sha256 61680e5c72e5… · 22339294997 bytes · M65
    • bielik-11b-v3.0-instruct mlx-bf16 · Not redistributed: ask the Room owner, or rebuild with scripts/02_convert_mlx.py at the pinned revisions · restricted · sha256 316155e1d979… · 22341216400 bytes · M66
    • models_manifest.json · https://github.com/stw2/tokenizer-science-tax/blob/2fc1a1b415a06b7648cdb37319ad573ccdc96a81/experiments/E02-digit-policy/results/models_manifest.json · public · sha256 3e07737b1f16… · 8931 bytes · M67
    • convert_mlx.log · Archived private repository, experiments/E2/results/convert_mlx.log at commit 707e7b3; not ported (a log with absolute paths); its facts are in this attempt's environment facts · unavailable · sha256 d7e17f8eb324… · 464 bytes · M68

    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.