← All runs

Evaluation run

self-correction chains

6 July 2026, 13:53 · eval-engine · 232478e · 89 cases

Did it hear the words right?

This run graded typed text only, so there is nothing here to have misheard.

Did it clean up correctly and safely?

96%
Light, ships on by default
85/89 cases passed · 231 ms typical
92%
Polish, experimental
82/89 cases passed · 265 ms typical
When a case fails, whose fault is it?
0 the microphone misheard the words  ·  11 the cleanup itself slipped

Every case

Each case shows the instant on-device cleanup, then what each mode did to it. Added words are marked, removed words are struck. For a spoken case, Heard is what the microphone transcribed and Was said is the reference it is scored against.

89 cases match

disf-scratchTypedclean
Instantthe deadline is friday actually no let's say next monday
LightThe deadline is friday actually no let's say next Monday. 175 ms
PolishThe deadline is friday actually no let's say next Monday. 182 ms
disf-selfcorrect-nameTypedclean
Instantsend it to john i mean jane
LightSend it to john i mean Jane. 152 ms
PolishSend it to john i mean Jane. 172 ms
disf-selfcorrect-timeTypedclean
Instantlet's meet at 4:00
LightLet's meet at 4:00. 229 ms
PolishLet's meet at 4:00. 269 ms
edge-emailTypedneeds review
Instantemail me@john.smith at gmail dot com
Lightemail me@john.smith at gmail dot com me@john.smith@gmail.com 213 ms
Polishemail me@john.smith at gmail dot com 222 msthe cleanup was rejected, so the original was kept
Why polish?the result is missing “john.smith@gmail.com”; the result still contains “dot com” and “ at gmail”, which this case forbids
edge-gibberishTypedclean
Instantasdf qwerty test test one two three
Lightasdf qwerty test test one two three. 220 ms
Polishasdf qwerty. Test. test One two three. 266 ms
edge-mixedTypedclean
Instanthey so the the lyzr demo is at 5:00 and accuracy is like 90% now
LightHey, so the the Lyzr demo is at 5:00 and accuracy is like 90%. now 449 ms
PolishHey, so the the Lyzr demo is at 5:00, and accuracy is like 90%. now 493 ms
edge-shortTypedclean
Instantyeah sounds good
LightYeah, sounds good. 154 ms
PolishYeah, sounds good. 184 ms
edge-singleTypedclean
Instantperfect
LightPerfect. 110 ms
PolishPerfect. 128 ms
faith-already-cleanTypedclean
InstantThe quarterly report is due on Friday.
LightThe quarterly report is due on Friday. 200 ms
PolishThe quarterly report is due on Friday. 215 ms
faith-codeTypedclean
Instantwrite a python function to reverse a string
LightWrite a python function to reverse a string. 218 ms
Polishwrite a python function to reverse a string 559 msthe cleanup was rejected, so the original was kept