Execution attempt

Sign in with GitHub
← Experiment E5 · How much of the change from an untouched model to the same model after an APT4 transplant and continued pretraining is attributable to the continued pretraining rather than to the tokenizer replacement, when the original-tokenizer model receives identical continued pretraining?

Execution attempt · A46 · planned

Scores the digit probe on the untouched base and arm A's five milestones.

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 @ b4c55e6e7d26917e239ccbe2f11886c4d4a733a2

Reference checked 2026-09-14 13:12 UTC. Later commits, branches or plan changes do not retarget this attempt.

Command
bash scripts/06_score_checkpoint.sh digits armA t0_base=Qwen/Qwen2.5-1.5B ckpt_00100M=ckpts/armA/ckpt_00100M ckpt_00200M=ckpts/armA/ckpt_00200M ckpt_00300M=ckpts/armA/ckpt_00300M ckpt_00400M=ckpts/armA/ckpt_00400M ckpt_00500M=ckpts/armA/ckpt_00500M
Working directory
experiments/E05-control-arm
Configuration paths
None
Parameters
300 items in each of three formats, 4-shot plain-text prompts, greedy decoding with batch 24, completion cut at the first newline, graded by E2's dual-locale grader after its 31-case golden suite; BOS parity checked per model.
Environment
3.12.13; PyTorch MPS, bf16; 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.

  • t0_base · checkpoint · M124 Qwen2.5-1.5B · checkpoint · Download
  • ckpt_00100M · checkpoint · M136 ckpt_00100M · raw output · Ask the reporter
  • ckpt_00200M · checkpoint · M137 ckpt_00200M · raw output · Ask the reporter
  • ckpt_00300M · checkpoint · M138 ckpt_00300M · raw output · Ask the reporter
  • ckpt_00400M · checkpoint · M139 ckpt_00400M · raw output · Ask the reporter
  • ckpt_00500M · checkpoint · M140 ckpt_00500M · raw output · Ask the reporter
  • probe_digits_spec.json · configuration · M150 probe_digits_spec.json · raw output · Download
  • probes.jsonl · evaluation data · M63 probes.jsonl · raw output · Download

Delivered events

  1. Registered

    #1

    Scores the digit probe on the untouched base and arm A's five milestones.

    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.

    • probe_digits_armA.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_armA.jsonl · public · sha256 7fc4a8c81894… · 5544 bytes · M154
    • probe_digits_armA.jsonl as originally written · Archived private repository, experiments/E5/results/probe_digits_armA.jsonl at commit 5389273; the public copy replaces absolute model paths with paths relative to the experiment folder · unavailable · sha256 db7b3fc337d9… · 5804 bytes · M155
    • armA/t0_base.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/t0_base.jsonl · public · sha256 90aa03cda12a… · 246394 bytes · M156
    • armA/ckpt_00100M.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/ckpt_00100M.jsonl · public · sha256 bba453af6c16… · 246479 bytes · M157
    • armA/ckpt_00100M.jsonl as originally written · Archived private repository, experiments/E5/results/probe_digits_raw/armA/ckpt_00100M.jsonl at commit 5389273; the public copy replaces absolute model paths with paths relative to the experiment folder · unavailable · sha256 887eda883ba7… · 246531 bytes · M158
    • armA/ckpt_00200M.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/ckpt_00200M.jsonl · public · sha256 9d058808dcac… · 246073 bytes · M159
    • armA/ckpt_00200M.jsonl as originally written · Archived private repository, experiments/E5/results/probe_digits_raw/armA/ckpt_00200M.jsonl at commit 5389273; the public copy replaces absolute model paths with paths relative to the experiment folder · unavailable · sha256 0d51ff639ab3… · 246125 bytes · M160
    • armA/ckpt_00300M.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/ckpt_00300M.jsonl · public · sha256 42bb01675fe3… · 246127 bytes · M161
    • armA/ckpt_00300M.jsonl as originally written · Archived private repository, experiments/E5/results/probe_digits_raw/armA/ckpt_00300M.jsonl at commit 5389273; the public copy replaces absolute model paths with paths relative to the experiment folder · unavailable · sha256 a4c46e22aa69… · 246179 bytes · M162
    • armA/ckpt_00400M.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/ckpt_00400M.jsonl · public · sha256 bee0fda00391… · 246381 bytes · M163
    • armA/ckpt_00400M.jsonl as originally written · Archived private repository, experiments/E5/results/probe_digits_raw/armA/ckpt_00400M.jsonl at commit 5389273; the public copy replaces absolute model paths with paths relative to the experiment folder · unavailable · sha256 7a39aa37b385… · 246433 bytes · M164
    • armA/ckpt_00500M.jsonl · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/ckpt_00500M.jsonl · public · sha256 adcb7e4d2317… · 246393 bytes · M165
    • armA/ckpt_00500M.jsonl as originally written · Archived private repository, experiments/E5/results/probe_digits_raw/armA/ckpt_00500M.jsonl at commit 5389273; the public copy replaces absolute model paths with paths relative to the experiment folder · unavailable · sha256 27126e2460a4… · 246445 bytes · M166
    • armA/prompt_dump_t0_base.json · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/prompt_dump_t0_base.json · public · sha256 da2e51f25db4… · 777 bytes · M167
    • armA/prompt_dump_ckpt_00100M.json · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/prompt_dump_ckpt_00100M.json · public · sha256 da2e51f25db4… · 777 bytes · M167
    • armA/prompt_dump_ckpt_00200M.json · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/prompt_dump_ckpt_00200M.json · public · sha256 da2e51f25db4… · 777 bytes · M167
    • armA/prompt_dump_ckpt_00300M.json · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/prompt_dump_ckpt_00300M.json · public · sha256 da2e51f25db4… · 777 bytes · M167
    • armA/prompt_dump_ckpt_00400M.json · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/prompt_dump_ckpt_00400M.json · public · sha256 da2e51f25db4… · 777 bytes · M167
    • armA/prompt_dump_ckpt_00500M.json · https://github.com/stw2/tokenizer-science-tax/blob/9a57aa40f4e774136f51b6730c33266d59094a45/experiments/E05-control-arm/results/probe_digits_raw/armA/prompt_dump_ckpt_00500M.json · public · sha256 da2e51f25db4… · 777 bytes · M167

    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.