Skip to content

packs: ⚡️ turn off Cerebras reasoning in the plain-English rewrite - #169

Merged
yasyf merged 1 commit into
mainfrom
plain-english-no-reasoning
Sep 25, 2026
Merged

yasyf merged 1 commit into
mainfrom
plain-english-no-reasoning

Conversation

@yasyf

@yasyf yasyf commented Sep 25, 2026 •

Copy link
Copy Markdown
Owner

Context: Follows #165. The plain-English rewrite calls Cerebras qwen-3.8-27b, which reasons by default. On a 2,300-character reply it spent a median 6,300 reasoning tokens for about 520 output tokens, so even with #165's 6-second cap many rewrites timed out and showed the original.

Summary: plain_english passes reasoning_effort="none" to OpenAiEndpointBackend, and the spawnllm pin moves to >=0.14.0,<0.15, the first release with that field (yasyf/spawnllm#10).

Motivation: The owner chose to make reasoning_effort a typed field in spawnllm rather than override the backend's request planning here.

Details: 20 calls through the real plain_english.plain_english() path, same prompt and harness, a fresh nonce per call:

p50 p95 max fell back to original
#165 (spawnllm 0.13.4, reasoning on) 4.42s 6.02s 6.47s 3 of 20 (6s cap)
this PR (spawnllm with reasoning_effort, "none") 0.52s 0.98s 1.13s 0 of 20

The "after" row ran against yasyf/spawnllm#10 installed from its branch. After v0.14.0 was published, 12 more calls against the locked release gave 0.52s p50 and 0.74s p95, with no fallbacks. uv lock --upgrade-package spawnllm moves the lock from 0.13.4 to 0.14.0. test_streamed_message_is_rewritten_once asserts that the backend carries reasoning_effort == "none". Both changelog entries move to Unreleased: #165's landed under 12.56.0, which was tagged before #165 merged. pytest tests/test_plain_english.py tests/test_plugin_hooks.py passes.

@yasyf
yasyf changed the base branch from plain-english-budget to main September 25, 2026 03:05
Context: Follows #165. The plain-English rewrite calls Cerebras `qwen-3.8-27b`, which reasons by default. On a 2,300-character reply it spent a median 6,300 reasoning tokens for about 520 output tokens, so even with #165's 6-second cap many rewrites timed out and showed the original.

Summary: `plain_english` passes `reasoning_effort="none"` to `OpenAiEndpointBackend`, and the spawnllm pin moves to `>=0.14.0,<0.15`, the first release with that field (yasyf/spawnllm#10).

Motivation: The owner chose to make `reasoning_effort` a typed field in spawnllm rather than override the backend's request planning here.

Details: 20 calls through the real `plain_english.plain_english()` path, same prompt and harness, a fresh nonce per call:

| | p50 | p95 | max | fell back to original |
|---|---|---|---|---|
| #165 (spawnllm 0.13.4, reasoning on) | 4.42s | 6.02s | 6.47s | 3 of 20 (6s cap) |
| this PR (spawnllm with reasoning_effort, `"none"`) | 0.52s | 0.98s | 1.13s | 0 of 20 |

The "after" row ran against yasyf/spawnllm#10 installed from its branch. After v0.14.0 was published, 12 more calls against the locked release gave 0.52s p50 and 0.74s p95, with no fallbacks. `uv lock --upgrade-package spawnllm` moves the lock from 0.13.4 to 0.14.0. `test_streamed_message_is_rewritten_once` asserts that the backend carries `reasoning_effort == "none"`. Both changelog entries move to Unreleased: #165's landed under 12.56.0, which was tagged before #165 merged. `pytest tests/test_plain_english.py tests/test_plugin_hooks.py` passes.
@yasyf
yasyf force-pushed the plain-english-no-reasoning branch from 6833470 to f1e5c14 Compare September 25, 2026 03:06
@socket-security

Copy link
Copy Markdown

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff Package Supply Chain
Security
Vulnerability Quality Maintenance License
Updatedpypi/​spawnllm@​0.13.4 ⏵ 0.14.096 +1100100100100

View full report

@yasyf
yasyf merged commit 631bf97 into main Sep 25, 2026
11 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant