Skip to main content

numerics-debugging

Capture and compare per-op activations between two TorchTitan runs to spot numerics divergence (eager vs aot_fx_trace, FSDP vs no-FSDP, before vs after a refactor). Use when the user wants to debug bitwise / numeric drift in training, or invokes /numerics_debugging.

Jump to install

Source facts

Repository
L27z18328742/torchtitan
Last source activity
August 28, 2026 at 16:39
Detected SKILL.md language
English
Stars
0
Forks
0

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

File Explorer
5 files

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
numerics_debugging
description
Capture and compare per-op activations between two TorchTitan runs to spot numerics divergence (eager vs aot_fx_trace, FSDP vs no-FSDP, before vs after a refactor). Use when the user wants to debug bitwise / numeric drift in training, or invokes /numerics_debugging.
# Numerics Debugging (DebugMode-based) Per-op activation capture + comparison toolkit. Captures activations on a designated step via `torch.utils._debug_mode.DebugMode`, then diffs two captures into an HTML report to surface numerics divergence between runs that should agree (bitwise, or within float32 reduction-order noise). Two pieces live in `scripts/`: - `activation_tracer.py` — runtime capture, driven from `Profiler` by `ActivationCaptureProfiler`. Output: `{dump_folder}/numerics/rank_{N}_activations.log`. - `compare_numerics.py` — diffs two logs, produces an HTML report. Standard library only, no torch import. They sit **outside** the torchtitan package on purpose. Nothing in core torchtitan or `graph_trainer` references them; an agent must edit torchtitan to wire the tracer in before a capture run, and revert the edits when done (they don't belong on `main`). Because `.claude` is not a valid Python package name, `activation_tracer` is imported by putting `scripts/` on `sys.path` rather than by dotted module path — see [references/patching.md](references/patching.md). > The two runs being compared **must use the same dtype and seed**. The > matcher keys on shape + float64 L1 norm; a precision change (bf16 vs > fp32) makes every row diverge and the matcher degrades to the > structural-only `stats` pass. ## Workflow 1. **Patch torchtitan** to wire the capture into `Profiler` (and `graph_trainer` if you're capturing the traced path). Full patch set: [references/patching.md](references/patching.md). 2. **Capture twice**, once per run you want to compare: ```bash ./run_train.sh \ --dump_folder ./outputs/run_A \ --training.steps 2 \ --profiler.dump_numerics \ --profiler.profile_freq 2 \ --debug.seed 42 \ --debug.deterministic \ --training.mixed_precision_param float32 ``` The capture step is `profile_freq`. With `profile_freq=2` and `training.steps=2`, step 1 warms up and step 2 is the snapshot. Capture adds ~10–40% memory only on the capture step (stats are computed inline in float64; tensors aren't held). 3. **Diff** the two logs: ```bash python .claude/skills/numerics_debugging/scripts/compare_numerics.py \ outputs/run_A/numerics/rank_0_activations.log \ outputs/run_B/numerics/rank_0_activations.log \ --name1 run_A --name2 run_B \ -o diff.html ``` Open `diff.html`. Each row pairs one op from each run; cells turn red when a stat diverges; the "Match method" chip shows which of the four matching passes (override / exact key / fuzzy key / stats) paired the row. ## Customizing Excluded ops, numel / dtype filter, hash function, the manual-override file format, and HTML appearance are all tunable. See [references/customization.md](references/customization.md), which also catalogs the common eager-vs-traced mismatch patterns (AC-recompute FQN drift, per-layer counter shifts, collective renaming) you'll see in the diff and how to express them as overrides.
View on GitHub