Repository navigation
0.70.2 P0: Qwen3.5-0.8B-IQ4_XS.gguf golden_output thinking_on: measure the thinking budget for this model (known failure in 0.70.1) #4663
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: Qwen3.5-0.8B-IQ4_XS.gguf
- Host(s): lambda
- Check: golden_output thinking_on
- 0.70.1 result: FAIL: no measured thinking budget (named contract refusal in contracts/thinking-budgets-v1.yaml; the GPU ran the model)
- v0.69.5-rc.1 result: FAIL: GPU forward fell back to CPU (the check passes on neither binary; 0.70.1 gets further)
- Logs: 0.70.1: /mnt/nvme-raid0/tmp/evid-0701b/lambda-ladder.log; rc.1: /mnt/nvme-raid0/tmp/screen-0701/logs/inv_Qwen3.5-0.8B-IQ4_XS.gguf.log, /mnt/nvme-raid0/tmp/screen-0701/receipts-0695/lambda.only-inv_Qwen3.5-0.8B-IQ4_XS.gguf.json
- Reproduce:
scripts/model_ladder.sh --host <lambda|gx10> --only inv:Qwen3.5-0.8B-IQ4_XS.gguf
Done when: the ladder row is OK on lambda and gx10.
Activity
evidence(pr-review): #4676 receipt for 2c2fd35 (independent reviewer)
fix(contracts): cite in-tree evidence for the IQ4_XS sweep; owed_by s…
fix(contracts): cite in-tree evidence for the Q4_K_M re-measurement (#…
noahgift commentedon Oct 6, 2026
B6 probe on lambda (RTX 4090),
apr 0.70.1 (6192aa9b88), sha256d8208cdfc6d3686a77db9e5c575c9ca14141eda9754f9b648b739cce8cc6d5aa, measured at 0.70.2's merge candidate 6192aa9.Backend:
cudaran rc 0 (GPU engaged).Result: still FAIL —
golden_output: golden_output_thinking_on: no measured thinking budget for Qwen3.5-0.8B-IQ4_XS.gguf (contracts/thinking-budgets-v1.yaml declares no budget for it; refusing rather than inheriting default). Same contract refusal as the 0.70.1 result cited in this issue.Receipt:
lambda.only-inv_Qwen3.5-0.8B-IQ4_XS.gguf.json(local to this run). Not yet OK on B6. gx10 leg still pending (build starts 09:15Z).noahgift commentedon Oct 6, 2026
Evidence row (B6 probe, GB10 sm_121 CUDA host): not_measured, not a close.
6192aa9b8), sha256787de0ca341342e9bd59502e3638ce99e05a0b839eceda52e8f6f56a9499eda7--only inv:Qwen3.5-0.8B-IQ4_XS.ggufQwen3.5-0.8B-IQ4_XS.ggufis not present there. The Qwen3.5-0.8B files present areQ4_K_M(see #4666) andUD-IQ2_XXS.A thinking-budget measurement for this model needs a host that holds the file, or the file staged here. No budget value can be read from this row.
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.
Agent: aprender-clerk