nils/jspace
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

pin gemma-4 model revisions (upstream 12B template regression: thought channel -> empty generations)

Browse Source 
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Nils
2026-07-14 17:13:35 +02:00
co-authored by Claude Fable 5
parent 378bb36b8c
commit e5fc453005
2 changed files with 18 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
LESSONS.md
+10
View File
@@ -76,3 +76,13 @@ reproduction: [`RESULTS.md`](RESULTS.md). Everything on `google/gemma-4-E2B-it`
3. **Verify with the lens, gate with the labels**: the J-lens picks the band,
measures whether loops compute, and diagnoses failures; STaR difficulty
labels supervise both the curriculum and (next) the adaptive-depth gate.
## 12. Pin model revisions on fresh nodes
Upstream updated google/gemma-4-12B-it mid-project: the new chat template
adds a `<|channel>thought` scaffold, and greedy generation closes the empty
thought channel and stops — every generation decodes to "". Symptom:
0/500 pass rates with rc=0 (looks like a harness bug, is a silent model
swap). E2B was unaffected. Fix: `hf download --revision <hash>` + repoint
`refs/main` in the cache; node_setup.sh now pins both models (12B
0e2b1058…, E2B 9dbdf8a8…). Rule: any cross-node result assumes identical
model revisions — pin them, don't trust "main".