test(kgemma): re-enable the Gemma 4 E2B tool-call smoke test — passes under engine 0.53.0 - #399
Merged
Merged
Conversation
… under engine 0.53.0 The test was @ignore'd because the real E2B checkpoint emitted coherent English on the calculator prompt but no <|tool_call> markup. Against the 0.53.0 engine (self-healing kernel dispatch, dense-FP32-from-any-storage, decode-shaped FP32 kernels) the Q4_K_M checkpoint now emits <|tool_call>call:calculator{expression:...} and every assertion holds in ~80 s. Still env-gated on GEMMA4_E2B_MODEL_PATH, so CI skips without the checkpoint. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Gemma4E2BToolCallSmokeTestwas@Ignored: the real E2B checkpoint emitted coherent English on the calculator prompt but no<|tool_call>markup (suspected prefill/sampling divergence). Verified locally against the published SKaiNET 0.53.0 withgemma-4-E2B-it-Q4_K_M.gguf: the model now emits<|tool_call>call:calculator{expression:...}and every assertion in the test holds (83 s). Re-enabling it as the regression guard; it stays env-gated onGEMMA4_E2B_MODEL_PATH, so CI without the checkpoint skips as before.Also verified in the same pass (no changes needed):
FunctionGemmaOfficialGgufTest3/3 (parsedget_weathercall) andQwenToolCallSmokeTest2/2 with Qwen3-1.7B-Q8 (well-formed<tool_call>calculator call).