Textbook order: collected advice, then his comment on how it applies to
LLMs, then the exercises. Content left for him to write.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Small is under a paragraph; large means work like comparing against a
reference repo. Do all applicable small ones, pick one large one.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Source is his own message of 2026-08-25, spelling fixed and slightly more polite as he
asked, with each mistake pointing at the exercise that answers it. Also adds his rule
that a job is never abandoned without doing the exercises, one at a time.
Two gaps the existing 13 did not cover, found by mining the evidence cache against
wassname's list of common AI-agent failures. Quotes are verbatim from
docs/evidence/ (Steinhardt, Rahtz, Nanda, Goodfellow-Bengio-Courville).
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Interpretations must read as evidence for or against a claim, never as a cause,
because a cause list gets picked from, called certain, and used to stop.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Both were promised by the description and absent from the procedure. P2 now
prints the formatted examples; P3 opens with the bug assumption.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
A sonnet subagent had ml-debug listed, with 'after a run finishes or crashes' in
the description, and did not invoke it for 'read the last pueue log'.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Sonnet routed to P1 correctly then declined it: 'the task only asked to read, not
to audit or act'. Reading a log is exactly when the measurement table gets filled.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
A sonnet subagent read the skill, summarised the ritual list, ran none of them,
and closed with 'everything checks out'. Phase triggers do not fire; a sentence
the agent watches itself write might.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Quoted whole from the 2017 thread, since the article author asked to merge it
and never did. Verified line by line against the thread cache. Covers sample
size from a cumulative-mean plot, KLD/Dice on unbalanced data, augmentation
bounded by feature std, dummy metrics, jumpy validation loss, testing the
framework itself, activation swaps, and loss-curve shapes.
102 of them, 48 already wrong after today's refetches. They rot every time a
cache is refetched and buy nothing the quote text does not: the caches are
verbatim, so the quote itself is the anchor. Descriptive labels stay.
All four came from the AI-generated deepwiki page, not the experiment log,
and one inverted the finding: the log says clipping was removed because
'Grad norm never exceeds 1.0 naturally', and the real distributed gotcha was
clipping local norms before sync. Footnote now points at dev/LOG.md on master
and at the renamed cache.
An /abs/ link is a stub: it costs a second lookup before anyone can check the
quote, and it is what let abstract-only caches look sourced. 77 links across
the skill files and 4 cache headers. Verbatim source bodies untouched, their
reference lists are the authors' text.
- SKILL.md: 3 new entries (exploration-over-exploitation + nuisance HPs,
test-set contamination, loss-spikes-mean-bad-data-pocket) and an Ng
100-misclassified-examples quote under inspect-the-data
- refs/llm_judges.md: position/verbosity/self-preference biases (Zheng,
Wang 66/80 flip, Panickssery) + mitigation checklist from verdict docs
- Lones pitfalls linked as the exhaustive 36-item do/don't checklist
- 6 new frozen evidence files; Hamel evals link in further reading
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Both quote-verbatim with frozen evidence: the 2018 tweet thread (mirrored
via threadreaderapp, x.com blocks fetching) slots after overfit-one-batch;
CACE (NIPS 2015, entanglement section transcribed from the PDF) gives
Always-Be-Ablating its why.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
SKILL.md is now folklore only: verbatim practitioner quotes ordered
most-general-first, transformer/LLM fine-tuning entries in their own
section, minimal context, links and footnotes. New sources: unsloth,
axolotl (+training stability), HF course ch8.4, Bekman debug_utils
(evidence frozen in docs/evidence/).
The synthesized material (mental models, priors, symptom tables, agent
loop, triage, anti-patterns) moves to PLAYBOOK.md, framed as menus of
hypotheses rather than authoritative diagnoses. Made-up symptom tables
no longer sit next to sourced quotes.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
NaN poisoning: inject NaN where info must not come from (future/test/labels), run the real pipeline, assert past outputs stay finite. Documents false negatives (pandas skipna, nanmean) and false positives (softmax rows, batch stats). Backprop-to-input is its gradient dual for inside the model; quote already frozen in docs/evidence/karpathy_recipe_training_nn_2019.md.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
- Promote the general (non-RL-specific) Spinning Up lessons up to the main
folklore: "broken code fails silently", "you can't tell it's broken if you
can't see that it's breaking", and test on more than one setup.
- Add gwern's "Unseeing" to the data theme: you can't read what you actually
wrote, hence fresh eyes / a fresh-eyes subagent.
- New "Research taste (adjacent to debugging)" section with verbatim quotes,
each cached: Neel Nanda (your research is false by default; excitement is
evidence of bullshit; read your data), Ulisse Mini (understand the system to
shrink the search space), John Wentworth (gears-level models are capital
investments vs cheap black boxes).
All quotes verbatim from cached sources; 25/25 footnotes resolve.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Gather debugging folklore from more practitioners, each a verbatim quote
checked against a cached source copy (footnoted with line numbers):
- koaning (Vincent Warmerdam), "Bad Labels": benchmark labels are often wrong;
find them with confidence-sorted errors.
- gwern, the tank-detection legend: the canonical data-leakage parable, plus
the scout-mindset twist that it's a likely-unsourced urban legend.
- Patrick Kidger, "Just Know Stuff": why research code is buggy ("kludge ...
bugs that don't cripple things only because some other bug stops them") and
"never accept the kludge". Plus a one-line jaxtyping pointer for shape bugs.
- nanochat (Karpathy): BOS-alignment fake metric improvement; all-ranks must
clip on inf (a multi-GPU bug single-GPU testing hides).
- cleanrl "37 Implementation Details of PPO" -> RL sub-skill, as the canonical
proof that reference-impl details (not ideas) decide whether PPO works.
Trim the lucidrains item to one quote (it had ballooned). Add wassname credit
+ companion-gist link. All 20 footnotes resolve.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
Phil Wang's x-transformers is the canonical "the fix is in the code, not the
paper" catalogue. Add a folklore item on the most debugging-relevant trick:
QK / cosine-sim normalization to stop attention logits overflowing (the usual
cause of transformer loss spikes/divergence), plus the BLOOM/YaLM
post-embedding LayerNorm. Two verbatim lucidrains quotes, footnoted to the repo
+ a cached README copy with line numbers. Doubles as the modern concrete
example for the read-a-working-implementation section.
Co-Authored-By: Claudypoo <288921227+claudypoo@users.noreply.github.com>
