th#1803 docs: bindings-as-config + tensor-layout contract - #657
Merged
Conversation
The authoring guide told authors how to quantize in setup(); Paul rejected that pattern (DESIGN-RULINGS §1.32). Quant selection is binding config, code is quant-generic over its declared tensor layout. report_applied_lane stays — it now covers the runtime-gated engagement half, and the two un-converted endpoints.
…ector is dead Paul's second and third same-day follow-ups. Quantization is ahead-of-time only (conversion endpoint) — boot-quant costs cold-boot seconds and wasted transfer, and the path is deleted rather than kept as a fallback. The flavor system is an arbitrary-string sub-selector being replaced by tensor-layout-contract compatibility (DESIGN-RULINGS §1.33; th#1809 hub, pgw#1143 SDK), so the flavor spellings in these docs are marked as current-state, not the target.
…or= is a dying axis
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Doc-only. Paul's 2026-08-11 rulings (tracker th#1803, DESIGN-RULINGS §1.32).
docs/endpoint-authoring.md— the section that taught authors how to quantize insetup()is retitled "Do NOT quantize weights insetup()": quant selection is binding config, code is quant-generic over its declared tensor layout.report_applied_lane(pgw#1104) is kept and rescoped — it still covers runtime-gated engagement (kernel/compile arms), and remains mandatory for the two shipped endpoints that predate the ruling. Also: the tensor-binding-contract section now says to declare the layout you execute, not a quant; thestorage_dtypeparagraph names flavor choice as deploy config.docs/compile-cache.md— the tensor-layout contract is a core platform feature; compatibility is answered ahead of time at rebind and request time.Companion PRs: tensorhub #1045, inference-endpoints
1803-docs.🤖 Generated with Claude Code