decoder-concept-page#211
Conversation
… for canonical concept discovery]
…coder concept-page slice]
|
Implemented This focused contract proves:
Local validation completed before push:
|
|
Addressed the failing mergeability blocker on the current PR head after build-export failed in CI. What changed:
Local validation on commit 14fb41a:
|
|
Addressed the failing mergeability blocker on the current PR head after failed in CI.\n\nWhat changed:\n- hardened so the header search probe waits for the visible , clears incidental page focus, and falls back to keyboard activation when Playwright times out on a pointer click\n- added focused unit coverage in for the normal click path, timeout fallback path, and non-timeout error passthrough\n\nLocal validation on commit :\n- bun test v1.3.13 (bf2e2cec)\n- [content-runtime] Removing stale generated Fumadocs bindings -> .source Creating an optimized production build ... Route (app) ○ (Static) prerendered as static content Wrote advanced Orama search bootstrap to /Users/abdifamily/work/learn-agent-factories/.claude/worktrees/decoder-concept-page/out/api/search, /Users/abdifamily/work/learn-agent-factories/.claude/worktrees/decoder-concept-page/out/api/search.vi, /Users/abdifamily/work/learn-agent-factories/.claude/worktrees/decoder-concept-page/out/api/search.ja. |
1 participant
{
"project": "Model Atlas — Decoder Canonical Concept Page",
"branchName": "decoder-concept-page",
"description": "Publish one canonical English
decoderconcept page, backed by the existingconcept.decoderregistry record and localized messages, so readers can understand what a decoder is, why decoder-only stacks fit next-token generation, and how to move through nearby GPT-style architecture and generation docs.","context": {
"customerAsk": "Customer ask alignment: the live queue has drifted below the requested 3-4 active worker floor because batch 027 no longer has three active execution lanes. Refill with one narrow, customer-visible concept page that strengthens the GPT-style reader path without colliding with the open attention, tokenizer, or training PR surfaces. Add the canonical English concept page for
decoderusing the existing publishedconcept.decoderregistry record on currentmain. Scope: create the canonical concept page plus colocatedmessages/en.jsonand any neededassets.json; classify it correctly with the project docs; wire aliases, tags, and related links so readers can move betweengpt-2,gpt-3,autoregressive-generation,decode,causal-attention,transformer, andencoder-decoder; and add only the focused validation needed for the touched page and discovery surfaces. The page should explain in simple language what a decoder is, why decoder-only stacks are suited to next-token generation, and how a decoder differs from an encoder or an encoder-decoder model without turning the page into a paper survey. Keep the slice English-only and avoid locale work, attention-family rewrites, or broad taxonomy churn beyond what is required to land the page cleanly. Acceptance criteria: thedecoderconcept page exists on currentmainas a canonical docs page, resolves against the existing registry record, supports GPT-family discovery paths, and focused touched checks pass.","problem": "The repository already ships the published
concept.decoderregistry record and a glossary page for decoder, but it does not yet ship the canonical concept page that explains decoder as the broad architecture idea across language, seq2seq, and other generation settings. Readers therefore have a gap between the short glossary definition and the more specific pages forautoregressive-generation,decode,causal-attention,transformer, andencoder-decoder, with no plain-language bridge that explains what a decoder does, why decoder-only stacks naturally support next-token generation, and how that differs from encoder and encoder-decoder layouts.","solution": "Publish a canonical
/docs/concepts/decoderpage using the standard concept-page contract, English-only message-driven content, and a colocatedassets.jsononly if one structured teaching asset is genuinely needed. Reuse the existingconcept.decoderregistry record, keep discovery metadata aligned with the broad concept, connect the page to the existing glossary entry and the nearby GPT-family, generation, and architecture surfaces already represented on currentmain, and add only the focused validation needed to prove route, message, registry, and discovery integrity for this narrow slice."},
"acceptanceCriteria": [
"A published canonical docs page exists for
decoderwithkind: \"concept\",registryId: \"concept.decoder\", English messages, and a colocatedassets.json.","The page follows the concept-page contract and docs-writing standards with one folded
openingSummary, plain-language explanation, distinct section jobs, and no page-meta or process language.","The page explains what a decoder does, what
decoder-onlymeans in modern language models, and why left-to-right masked attention makes decoder stacks a natural fit for next-token generation.","The page clearly distinguishes decoder from encoder and from encoder-decoder systems without turning into a broad historical survey or paper roundup.",
"Registry-backed discovery metadata and relationships let readers move between the canonical
decoderpage and the touchedgpt-2,gpt-3,autoregressive-generation,decode,causal-attention,transformer, andencoder-decodersurfaces on currentmain.","The implementation remains English-only and avoids unrelated locale, attention-family, tokenizer, or broad taxonomy churn.",
"Quality gate: typecheck, lint, and focused tests pass."
],
"userStories": [
{
"id": "decoder-concept-page-001",
"title": "Align the existing decoder record for canonical concept discovery",
"description": "As a reader searching for decoders in GPT-style systems, I want the existing
concept.decoderregistry record to behave like a first-class broad concept so discovery surfaces route me to one canonical explainer instead of only the glossary entry or scattered model mentions.","acceptanceCriteria": [
"The existing
concept.decoderrecord remains the canonical backing record and is updated only as needed with aliases, tags, citations, or related ids that fit a broad decoder concept page.","Registry relationships connect the concept page to shipped nearby docs for
autoregressive-generation,decode,causal-attention,transformer, andencoder-decoder, while preserving the glossary bridge forglossary/decoder.","Representative GPT-family discovery surfaces already published on current
main, includinggpt-3and the current canonicalgpt-2surface if it exists as a model or paper-adjacent page, can point readers toward the canonical decoder explainer without creating new GPT-family pages in this slice.","Discovery metadata distinguishes the broad
decoderconcept page from the existing glossary entry without introducing duplicate canonical targets or broad taxonomy churn.","Typecheck passes",
"Tests pass"
],
"priority": 1,
"passes": true,
"notes": ""
},
{
"id": "decoder-concept-page-002",
"title": "Publish the canonical decoder concept page",
"description": "As a technical layperson learning how GPT-style models generate text, I want a dedicated
decoderconcept page so I can understand what the decoder stack does and why decoder-only models naturally support next-token generation.","acceptanceCriteria": [
"A canonical concept page exists at
/docs/concepts/decoderwith matching frontmatter,messages/en.json, and a colocatedassets.jsonthat is empty or minimal unless one decoder-oriented teaching asset is genuinely needed.","The page opens with one folded
openingSummaryand explains decoder in plain language before narrowing into GPT-style transformer usage.","The page clearly explains that a decoder turns internal context into output predictions, and that in decoder-only language models the same stack repeatedly predicts the next token from the tokens already seen.",
"The page clearly explains why causal or left-to-right attention makes decoder-only stacks suited to autoregressive generation.",
"The page clearly explains how encoder-decoder systems use a decoder differently from decoder-only stacks, and why encoder-only systems are usually built for representation or understanding rather than next-token output loops.",
"The page remains understandable in isolation for a first-time reader and complements the glossary entry rather than duplicating it word for word.",
"Typecheck passes",
"Tests pass",
"Verify in browser using the Browser plugin"
],
"priority": 2,
"passes": true,
"notes": ""
},
{
"id": "decoder-concept-page-003",
"title": "Route readers through the GPT-style decoder path",
"description": "As a reader moving through GPT-family architecture and generation topics, I want related docs, tags, and nearby page surfaces to guide me into the broad
decoderexplainer and onward to the right adjacent concepts.","acceptanceCriteria": [
"Representative discovery paths from
gpt-3, the current canonicalgpt-2surface onmain,autoregressive-generation,decode,causal-attention,transformer, andencoder-decoderexpose the canonicaldecoderconcept page as a reachable related destination where those surfaces are already published.","The concept page renders registry-backed related docs and tag surfaces that connect it to the named neighboring pages without hand-maintained duplicate navigation lists.",
"At least one touched neighboring discovery surface presents the new decoder page in a way a reviewer can follow without typing the exact concept slug directly.",
"Browser-visible rendering shows the title, folded summary, tags, and related-doc links without broken links or missing-content placeholders.",
"Typecheck passes",
"Tests pass",
"Verify in browser using the Browser plugin"
],
"priority": 3,
"passes": true,
"notes": ""
},
{
"id": "decoder-concept-page-004",
"title": "Add focused validation for the decoder concept-page slice",
"description": "As a maintainer, I want targeted automated proof for the
decoderconcept-page slice so route, message, registry, and discovery regressions are caught without unrelated test expansion.","acceptanceCriteria": [
"Validation or tests confirm the
/docs/concepts/decoderroute,concept.decoderrecord, and default English messages resolve together.","Coverage asserts at least one page-specific discovery outcome and at least one page-specific related-link or glossary-bridge expectation for the new decoder concept page.",
"Focused checks stay limited to touched content and discovery integrity rather than inventory snapshots, locale churn, attention-family rewrites, or unrelated suite expansion.",
"Typecheck passes",
"Tests pass"
],
"priority": 4,
"passes": true,
"notes": ""
}
]
}