gparse-library-inference
The groupant beside gparse/src/lib.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 |
| cost | symbolic source derived | invariant |
| descriptant | present | descriptant |
| invariant | partial | invariant |
| tests | partial 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 |
layout_transport | layout_transport | partial source derived model | layout_transport |
score_branch | score_branch | partial source derived model | score_branch |
label_rewrite | label_rewrite | partial source derived model | label_rewrite |
environment_gate | environment_gate | partial source derived model | environment_gate |
Model text
descriptant
Source-bound description
Nonempty plans forward, transpose arcs, decode Eisner, gather heads/tags/relations and scatter into original order.
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
Nonempty plans forward, transpose arcs, decode Eisner, gather heads/tags/relations and scatter into original order.
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
Empty slots remain empty/zero; root labels rewrite after scores are obtained.
Cost
Encoder O(BTd+L*(BTd²+BT²d+BTdf)) plus arc projections/scoring, relation/tag projections and sum n³ decoder; attention BH*T², transfer BT². Parent cost states each independent projection term.
Allocation, finite arithmetic, platform dimensions and external failures are not universally certified.
models
Typed model catalog
behavior_complement
Empty slots cannot receive a word prediction; root labels override raw relation argmax.
reverse_reconstruction
Output trees admit score cones and many vocabulary/hash antecedents.
layout_transport
lp(k,d,h) copied to sc[h][d]; decoded h flattened at k*T+h.
score_branch
scored=false avoids tag/relation log reductions but shares argmax/decoder.
label_rewrite
h=0 forces root; h!=0 with predicted root becomes dep.
environment_gate
GP_FUSED_SOFTMAX only0/1; parse before plan loop even for empty list.
tests
Checks and limits
Selected parent checks relevant to this fragment (not exhaustive): synthetic_cpu_prediction_preserves_slots_and_fallback (actual predict and predict_scored); load/load_vars remain source-modeled, no model-file runtime claim
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.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/gparse/src/lib.rs.groupant/inference.groupant/