Feature #3516
Fires on “nums” (not labeled yet: read the examples)
- Fires on
- 0.60% of tokens
- Mostly in
- agent
- Layer
- after block 4
Where it fires
- agent 46%
- python 45%
- worksheet 5%
- story 2%
- chat 2%
Makes the model say
nums exclamation tests typo`old "!"`.Fixed
Makes it not say
needprint not FromHello Now return hiding
Strongest examples
On text the dictionary never trained on. The underlined token is where it fires most.
-
python 1.96s)`. The tests pass now.<|assistant_end|><|bos|>`double_all(nums)` doubles every
-
python 1.96Leo!⏎5.75⏎```<|assistant_end|><|bos|>`evens(nums)` keeps only
-
python 1.9.py`, line 1.<|assistant_end|><|bos|>`keep_even(nums)` keeps only
-
agent 1.9<|tool_end|><|assistant_start|>Added `evens(nums)` to `listlib
-
python 1.89larger(9, 7))⏎print(sub_one(COUNT))⏎⏎#
-
python 1.85_of⏎⏎NUMBER = -9⏎print(magnitude(NUMBER))⏎print(
-
python 1.85(-2)` returns `False`.⏎<|bos|>`average_list(nums)` returns the average
-
python 1.84`return min(nums)`.⏎⏎```
-
agent 1.84newString=print(doubled(SCORES))⏎⏎⏎def
-
python 1.84`return sum(nums) / len(nums)`.⏎⏎```
-
python 1.82of a and b. For example, `plus(12, 3)` returns `
-
python 1.81Hello, Lia!⏎```<|assistant_end|><|bos|>`average_list(nums)` returns the average