Elamur Groupantsgparse-test-training-speed
The groupant beside gparse/tests/training_speed.rs in the Mova repository: what the code computes, what it never does, how to go back from its outputs, its invariants, cost and checks.
Facts
Implementation material. A Mova code model describes this code at one source revision: partial models, finite checks and their stated scope. It is not a reviewed general exposition, and its labels are shown as written, never raised.
Components
| component | label | where |
|---|---|---|
| approximant | partial | approximant |
| ast | partial | syntax.json |
| composition | partial | composition |
| cost | symbolic source derived | invariant |
| descriptant | present | descriptant |
| evidence | executed finite checks | evidence |
| invariant | partial | invariant |
| tests | executed bounded checks | tests |
Obernants
The typed relations around the forward model: the behaviour complement (what the code never does), the reverse reconstruction (every input compatible with an output) and specialised views.
| id | type | label | where |
|---|---|---|---|
behavior_complement | behavior_complement | partial source derived model | behavior_complement |
reverse_reconstruction | reverse_reconstruction | partial source derived model | reverse_reconstruction |
optimizer_effect | optimizer_effect | partial source derived model | optimizer_effect |
effect_boundary | effect_boundary | partial source derived model | effect_boundary |
assurance_scope | assurance_scope | partial source derived model | assurance_scope |
failure_precedence | failure_precedence | partial source derived model | failure_precedence |
Parts
Model text
descriptant
Source-bound description
Actual step helper builds arc/rel/tag cross-entropies, loss=arc+rel+0.5*(xpos+upos), backpropagates, optionally checks finite positive norms of word/QKV/arc/rel weights, calls AdamW step and returns pre-update loss. New tiny seeded CUDA single-step test checks positive finite loss, changed arc_w and finite updated tensor. Existing measured_optimizer_steps requires real model/corpus and is not run.
Epoch 2026-10-31
Prospective model was recorded before edits. Added tests use bounded synthetic fixtures; production algorithms unchanged. Coverage is partial, not proof. No trained corpus, live teacher or throughput experiment was executed.
approximant
Forward approximant
Actual step helper builds arc/rel/tag cross-entropies, loss=arc+rel+0.5*(xpos+upos), backpropagates, optionally checks finite positive norms of word/QKV/arc/rel weights, calls AdamW step and returns pre-update loss. New tiny seeded CUDA single-step test checks positive finite loss, changed arc_w and finite updated tensor. Existing measured_optimizer_steps requires real model/corpus and is not run.
Domain: valid source-specific dimensions and resources stated in invariant; no automatic production substitution. Numerical observations are finite checks, not an end-to-end relative bound. Executable Rust surrogate: unsupported gap; existing exact tests are not surrogates. Future metric-specific caps may be 1/2/5/10%, never above 10%.
invariant
Invariant and cost
Only ephemeral VarMap/optimizer state changes. One changed tensor does not certify all optimizer moments or convergence; no throughput claim. Original experiment permits <=128 sentences of lengths24..40, 88 steps and trajectory delta<0.005*scale, but excluded.
Cost
One bounded tiny Net step with O(parameters) AdamW state; existing experiment much larger and not authorized/executed.
Allocation, finite arithmetic, platform dimensions and external failures are not universally certified.
composition
Composition
Actual step helper builds arc/rel/tag cross-entropies, loss=arc+rel+0.5*(xpos+upos), backpropagates, optionally checks finite positive norms of word/QKV/arc/rel weights, calls AdamW step and returns pre-update loss. New tiny seeded CUDA single-step test checks positive finite loss, changed arc_w and finite updated tensor. Existing measured_optimizer_steps requires real model/corpus and is not run.
Scoped children
The byte-scoped children project distinct internal relations through the parent syntax file. The parent combines them in actual source order. Remaining declarations/tests outside child ranges belong to this parent. Their correctness does not establish external caller preconditions.
Interfaces and limits
Only ephemeral VarMap/optimizer state changes. One changed tensor does not certify all optimizer moments or convergence; no throughput claim. Original experiment permits <=128 sentences of lengths24..40, 88 steps and trajectory delta<0.005*scale, but excluded.
Given returned loss, all pre-update logits with same cross-entropy and compatible labels are antecedents; post-step weights cannot recover moments or gradients without optimizer state.
models
Typed model catalog
behavior_complement
Passing selected assertions excludes their stated counterexamples within the fixtures. No real-model/corpus optimizer experiment was executed.
reverse_reconstruction
Given returned loss, all pre-update logits with same cross-entropy and compatible labels are antecedents; post-step weights cannot recover moments or gradients without optimizer state.
optimizer_effect
Inspect happens before opt.step; failure in norm assertion prevents update. Returned loss is computed before update.
effect_boundary
Selected test process owns local fixture tensors/state; explicit CUDA tests serialize. No teacher/API, dataset or external model reads selected.
assurance_scope
Compiled selected assertions establish only finite observations; ignored experiments and backend allocation failures remain gaps.
failure_precedence
Tensor/loss/backward failures precede optimizer update; inspect assertions also precede update.
tests
Checks and limits
synthetic_single_optimizer_step
Selected commands run twice debug and twice release; exact receipts and source-to-test mapping retained in wave30 gparse external run directory. Tests do not certify allocation failure, all NaN/overflow inputs, trained-model accuracy or corpus timing. Compiled mutants cover selected boundaries; per-source coverage is recorded in the external map, not presumed.
See source execution evidence for exact names and unexecuted paths.
evidence
Execution evidence — 2026-10-31
Source: gparse/tests/training_speed.rs
- synthetic_single_optimizer_step: twice debug, twice release, all four exits 0.
Controls
1 source-local compiled assertion mutants; 1 observed control tests including dependency mutations. Mutant compile exit0, assertion exit101, byte-exact restoration and positive rerun exit0. Per-source controls are listed in external source-test-map.json, not implied for every source.
Limits
Only synthetic_single_optimizer_step using actual step helper; measured_optimizer_steps explicitly not run.
CUDA softmax comparisons use absolute tolerance2e-6. GELU grid maxima observed: values0.0004734993, derivatives0.00086838007, finite difference0.0000005534244635896357; respective strict caps5e-4/1e-3/1e-5. These are finite observations, not percentage profiles or throughput claims.
Receipts: external wave30 gparse run native-commands.json, results.json, controls.json and source-test-map.json. Sorted exact test names/outcomes match per profile; numeric diagnostics and timings need not be byte-identical. Strict scoped library Clippy reports ten pre-existing production lints; appended test blocks were formatted independently without production reformatting.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/gparse/tests/training_speed.rs.groupant/