Repository navigation
0.70.2 P0: Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf cuda serve /api/chat: gibberish output (known failure in 0.70.1) #4662
Description
Known failure shipped in 0.70.1 (aprender 0.70.1, candidate cc4463f) under C272 5b; P0 for 0.70.2. Not a regression: it fails the same way on v0.69.5-rc.1 (5660b08).
- Model: Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf
- Host(s): lambda, gx10
- Check: cuda serve /api/chat
- 0.70.1 result: FAIL: gibberish (' What is 2+2?' repeats), stream=false and stream=true
- v0.69.5-rc.1 result: FAIL: identical (lambda and gx10)
- Logs: 0.70.1 lambda: /mnt/nvme-raid0/tmp/evid-0701b/lambda-ladder.log; 0.70.1 gx10: gx10:~/.cache/aprender-0701/progress2.log; rc.1 lambda: /mnt/nvme-raid0/tmp/screen-0701/logs/old-inv_Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf.log; rc.1 gx10: /mnt/nvme-raid0/tmp/gx10-0701/rc1/ladder-only-inv_Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf.log
- Reproduce:
scripts/model_ladder.sh --host <lambda|gx10> --only inv:Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf - Related: qwen3moe (Qwen3-30B-A3B, Qwen3-Coder-30B-A3B, Q4_K_M) has NO working CUDA path: apr run --gpu falls back to CPU (rc 14), apr qa emits no gates — blocks 0.69.1 under the all-Q4_K-CUDA rule #3714 (describes a CPU fallback for qwen3moe; this is gibberish output from the CUDA serve route, a different symptom)
Done when: the ladder row is OK on lambda and gx10.
Activity
fix(serve): GGUF servers tokenize with the canonical byte-level BPE (…
noahgift commentedon Oct 7, 2026
0.70.2 ships on a CRUX-smoke emergency scope (operator ruling C322, A1 and A2). This issue is refused for 0.70.2 and moves to milestone 0.71.0, with no per-issue contract edit. The 0.70.2 release notes name it as a known failure.
noahgift commentedon Oct 14, 2026
P0 to P1 (2026-10-14 P0 triage): P0 now means it blocks the next release or carries the operator's own words on it; this issue does neither, so it stays on the 0.71.0 milestone at P1. Its 0.70.2 label belonged to a release that has shipped.
noahgift commentedon Oct 14, 2026
Re-parent: former parent #4689 is closed. Moved under open epic #3998 (same 0.71 train).
noahgift commentedon Oct 14, 2026
Re-parent: grouped under #4701 (the model-ladder ticket, open, under E1 #3998) with the other ladder known-fail rows #4661-#4664; same 0.71 train.
noahgift commentedon Oct 14, 2026
Folded into #4701 as a checklist row (TRIAGE-100, 2026-10-14): the open-ticket cap is 100, so this ticket is now tracked on that master and closed as not planned. Nothing is lost: reopen it when its work resumes, and an open PR for it reopens it. Pushed branch:
fix/4662-moe-serve-runon.Agent: aprender-clerk