[serge] integration failure triage - 2026-09-16

[github-actions]

Automated integration-failure triage for the daily CI window 2026-09-10 → 2026-09-16.

This issue was generated by AI-assisted automation. The grouping, summaries, and recommended follow-up can be incomplete or misleading; verify the failures before acting.

Serge dispatched one task per failure group below — each opens or updates its own PR on a serge/fix/itf-<fingerprint> branch. This table is refreshed in place as Serge runs: a group links its #<pr> when opened, shows 🚫 no fix when Serge found no safe change, ⚠️ task failed on error, or (pending) while still running (a late PR links on the next nightly run).

Dispatched failure groups

Model	Error	Occurrences	PR
qwen3_omni_moe (regressed by PR #48714)	mixed — other (2)	2	⚠️ task failed
generation	cuda_runtime — other (12)	12	🚫 no fix
deepseek_vl	other — other (3)	3	#48536
cwm	OOM — other (2)	2	🚫 no fix
eomt_dinov3	output_mismatch — tensor values differ (2)	2	#48882

Not dispatched — environment / dependency

These groups were triaged but NOT handed to Serge: their failure mode is a property of the runner or the environment, so no minimal source patch can fix them. They need a human (runner capacity, a dependency pin).

2 models ran out of device memory (7 failures) — needs runner capacity, not a patch, so none of these were dispatched: bamba (4), llama4 (3).

5 group(s) skipped — stopped failing before this run (2 integration tests regressed by commit 774a675d827e (PR #47827), edgetam, depth_anything, glm4v_moe +1 more). Not dispatched and not awaiting a human.

Outcome recap

Why each group that opened no PR ended without one, and what it cost — surfaced here from the Serge dashboard. Failing tests links straight to each test's dashboard view, since these are the groups a human has to pick up.

Group	Reason	LLM	Tokens (in / out)	Failing tests
qwen3_omni_moe (regressed by PR #48714)	task runner exited without reporting (exit code 1)	moonshotai/Kimi-K2.7-Code	2,092,722 / 3,697	test_small_model_integration_test_batch_audio_matches_single · test_small_model_integration_test_w_audio
generation	[not_reproduced] GPU reproduce: the targeted tests did NOT fail at the base commit (not_reproduced) — the failure is stale, already fixed, flaky, or environment-specific. Skipping this group; no investigation, no PR.	moonshotai/Kimi-K2.7-Code	— / —	test_validate_assistant · test_matches_immediate_check_at_max_length · test_matches_immediate_check_on_a_batch · +3 more
cwm	[not_reproduced] GPU reproduce: the targeted tests did NOT fail at the base commit (not_reproduced) — the failure is stale, already fixed, flaky, or environment-specific. Skipping this group; no investigation, no PR.	moonshotai/Kimi-K2.7-Code	— / —	test_cwm_generation_20_tokens · test_cwm_sliding_window_long_sequence

Generated 2026-09-17T00:12:11+00:00.