docs(blackwell): cascade post-mortem — 8 PRs / 7 defects / 1 root cause#1824
Merged
Conversation
Capture the institutional knowledge from the PMAT-698e..m + PMAT-700-B cascade that unblocked Phase 3 distillation training on Blackwell GB10. Highlights: - 5 lessons (symptom-similarity signal, macro audit, pre-warm contract halves, Blackwell exposes latent fragility, smoke-contract validity) - 3 interventions that would have caught the root cause in 1 PR: diagnostic logging from start; property test for pre-warm coverage; cross-architecture (sm_89 + sm_121) CI gate - Effort accounting: ~50 lines of net production code, ~250 lines of comments + spec across 8 PRs This post-mortem is upstream-shareable — recommend forwarding to the trueno team to inform trueno#200 (the official Blackwell JIT fix) plus any future cross-architecture work. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
noahgift
added a commit
that referenced
this pull request
May 20, 2026
2026-05-20 — real distillation 1.5B teacher → 0.5B student on
Blackwell GB10 with the full PMAT-698e..n + PMAT-700-B cascade active.
initial_loss = 7.6746
final_loss = 7.2036 ← LESS THAN initial
62 steps, 122.7s, no errors
F-DISTILL-SMOKE-001 ("final_loss < initial_loss") discharged.
Phase 3 of SPEC-DISTILL-001 is COMPLETE.
Evidence:
- evidence/distill-phase-3-real-kd/dispatch.json — dispatch manifest
- evidence/distill-phase-3-real-kd/launch-final-pass.txt — full training log
Run dir on gx10: /home/noah/runs/distill-smoke-20260520-070404/
Trained student checkpoint: student-trained.apr/model.safetensors
Cascade summary (all merged):
- #1804 PMAT-700-B (cuBLAS prewarm skip)
- #1808 PMAT-698e (workspace cap)
- #1809 PMAT-698f (APR magic in weights loader)
- #1810 PMAT-698g (non-LoRA backward pre-warm)
- #1813 PMAT-698h (rms_norm_gamma_reduce pre-warm)
- #1815 PMAT-698i (FWD-CACHE diagnostic logging)
- #1817 PMAT-698j (THE root cause — warm! macro key)
- #1820 PMAT-698k (cache-key alignment: rope fwd + rmsnorm eps)
- #1823 PMAT-698m (smoke setup: non-degenerate batch)
- #1824 (post-mortem doc)
- #1827 PMAT-698n (rmsnorm pre-warm at both 1e-6 + 1e-5 eps)
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Institutional-knowledge capture from the PMAT-698e..m + PMAT-700-B cascade that unblocked Phase 3 distillation training on Blackwell GB10 (sm_121). One session, 9 PRs, ~7h debugging.
Highlights
5 lessons:
\$keymust be substituted into the call, not just passed in.3 interventions that would have caught the root cause in 1 PR instead of 7:
get_or_compilefrom day one (one eprintln)pre_warm_keys(config) ⊇ runtime_keys(config)for any forward passEffort: ~50 lines net production code; ~250 lines comments + spec; 8 PRs.
Cascade table
Recommended distribution
This post-mortem is upstream-shareable — share with the trueno team to inform trueno#200 (the official Blackwell JIT fix) plus any future cross-architecture validation work.
Test plan
Docs-only PR; no behavioral changes.
🤖 Generated with Claude Code