Repository navigation
docs: add local decisions blog post and generated images - #56
Merged
Merged
Conversation
Adds a technical, code-sample-driven blog post for the v0.22.0 ElBruno.LocalLLMs.Decisions release, plus a companion assets file and three generated illustrations. All figures in the post come from live runs rather than estimates: routing confidences and latencies from the LocalDecisions sample, and the calibration-clamp and fp16 variance tables from docs/decisions-guide.md. The companion file records the source for each number so they can be re-verified. The post covers the honest calibration caveats as well as the happy path: the clamped choice:11+ bucket still reports saturated confidence, so ranking is usable but magnitude is not. Co-authored-by: Copilot App <223556219+Copilot@users.noreply.github.com> Copilot-Session: c71e4170-a73a-4ac2-8c59-4b80834e7968
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds a technical, code-sample-driven blog post for the v0.22.0
ElBruno.LocalLLMs.Decisionsrelease, ready to publish on elbruno.com.What's here
docs/blog/blog-local-decisions.mddocs/blog/blog-local-decisions-assets.mdimages/blog-local-decisions.pngdocs/blog/assets/local-decisions/one-forward-pass.pngdocs/blog/assets/local-decisions/calibration-clamp.pngEvery number is from a live run
No estimated or invented benchmarks. The companion file records the source for each figure so they can be re-verified:
dotnet run --project src/samples/LocalDecisions.docs/decisions-guide.md.Verification done
Build succeeded, 0 Error(s)— rather than eyeballed. This confirmed the parameterlessLayaOnnxDecisionClient()ctor,result.Probability, andresult.CalibrationClamped.LocalDecisionssample.docs/blog/.Note on the branch
This branch was previously squash-merged for v0.22.0, so git no longer saw
mainas an ancestor — a PR from it would have re-proposed all 80 already-merged files. I reset the branch ontoorigin/mainfirst, so this PR contains exactly the 5 new files.Tone
Follows the existing
blog-bitnet-launch.md/blog-inkling-support.mdconventions, and keeps the honest-caveats section: the clampedchoice:11+bucket still reports saturated confidence, so ranking is usable but magnitude is not, andChoiceOrNull(threshold)will never abstain there.Docs-only — no code or workflow changes.