Model card: the hook's 20-second limit

README.md

@@ -26,8 +26,10 @@ $ commit-suggestor download
 $ git add -p && commit-suggestor suggest
 ```
 
-`commit-suggestor install-hook` pre-fills the message on `git commit`, with an
-8-second limit and never a download inside a commit. The prompt is
+`commit-suggestor install-hook` pre-fills the message on `git commit`, with a
+20-second limit (`COMMIT_SUGGESTOR_TIMEOUT`) and never a download inside a
+commit. The hook loads the model afresh for each commit, so it waits several
+seconds on a Mac and more on a CPU. The prompt is
 `commit_suggestor/prompt.py` (prompt version 2, without repository context);
 asking the model any other way is testing a prompt it was not trained on.
 Subjects over 72 characters are shortened at a clause boundary
@@ -107,10 +109,10 @@ rule failed a check against human ratings in v1, so treat it as a proxy.
   same subject as fp16 in 25.11% [21.33, 29.11];
   ROUGE-L against fp16 2.27 [1.16, 3.41] points.
   7.73 s per subject on 4 Kaggle CPU cores,
-  close to the hook's 8-second limit: on a slow CPU, raise it with
-  `COMMIT_SUGGESTOR_TIMEOUT`.
+  after loading. On a slow CPU the hook can pass its 20-second limit; raise it
+  with `COMMIT_SUGGESTOR_TIMEOUT`.
 - MLX on an Apple M4, 16 GB: 2.68 s to load,
-  0.433 s per subject (p95 0.978 s),
+  then 0.433 s per subject (p95 0.978 s),
   1,446 MB peak memory.
 
 ## Limitations