fix(tensorzero): override Z.AI GLM-5 temperature to 0.1 for coding variants (SUPERSEDED) - #1510
POWERFULMOVES wants to merge 1 commit into
Conversation
…riants Per 4090 handoff report 2026-05-14 (PR #1508, item #1 in research/HANDOFF_4090_FIXES_MAY2026.md) and Z.AI coding plan docs: GLM-5-Turbo defaults to temperature=1.0 which is too creative for code generation and tool calling. Override to 0.1 (in the documented 0.0-0.3 coding range) for deterministic-enough generation while preserving small variation for retry diversity. Applied to 6 TensorZero variants that route to the Z.AI coding endpoint (`https://api.z.ai/api/coding/paas/v4`): functions.agent_zero (KiloCode primary): - hosted_zai_turbo (chat_zai_glm5_turbo, w=0.6) - hosted_zai_vision_turbo (chat_zai_glm5v_turbo, w=0.3) functions.coding_glm (code-gen specialist): - cloud_zai_turbo (chat_zai_glm5_turbo, w=0.8) - cloud_zai_vision_turbo (chat_zai_glm5v_turbo, w=0.4) - cloud_zai_glm51 (chat_zai_glm51) - cloud_zai_glm_flash (chat_zai_glm) Rationale comments added at both variant-block headers cross-referencing research/HANDOFF_4090_FIXES_MAY2026.md. Note (out of scope): `python3 -c 'import tomllib; tomllib.load(...)'` flags a pre-existing duplicate model declaration at line 549 (`models.ollama_spark_gemma4_31b_fp16` declared twice). Reproduces on origin/main; not introduced by this PR. TensorZero's Rust TOML parser is likely more permissive than Python's strict tomllib, hence the config probably loads fine at runtime. Worth a separate dedup PR. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
Warning Rate limit exceeded
You’ve run out of usage credits. Purchase more in the billing tab. ⌛ How to resolve this issue?After the wait time has elapsed, a review can be triggered using the We recommend that you space out your commits to avoid hitting the rate limit. 🚦 How do rate limits work?CodeRabbit enforces hourly rate limits for each developer per organization. Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout. Please see our FAQ for further information. ℹ️ Review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (1)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
POWERFULMOVES
changed the title
1 participant
Superseded by #1511 which contains the identical
temperature = 0.1additions totensorzero.tomlfor all 6 Z.AI coding variants, plus the ClawZ submodule bump. Merging #1510 first would create a conflict on #1511. Close this and merge #1511 instead.