CI, tokenizer fix, and the --bench throughput harness#5
Merged
Conversation
Compiles the release binary and runs the kernel unit tests on every push.
Builds a Qwen2.5-0.5B-shaped model with random weights (timing depends on matrix sizes, not values) and times decode throughput per quant scheme. CI runs it, so the forward pass is exercised end-to-end on every push.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Completes the project: green CI plus a benchmark harness, so the whole thing is verified to compile, test, and run.
What landed
cargo build --release+cargo teston every push (Ubuntu runner). All 13 kernel unit tests pass; the release binary builds clean.onig) so it compiles (needs a C toolchain, present on Linux/macOS/CI).--bench— throughput on a randomly-weighted Qwen2.5-0.5B-shaped model (no weights needed); CI runs it, so the forward pass is exercised end-to-end.Measured (2-vCPU CI runner)
Quantization is a 4x/7x memory win; the scalar dequant trades throughput (SIMD dequant is the next optimization). The engine now compiles, tests green, and runs on real hardware.