Skip to content
mini-lab
Log in Sign up
Features of prelude-2

Feature #1938

Fires on “_” (not labeled yet: read the examples)

Fires on
0.42% of tokens
Mostly in
agent
Layer
after block 4

Where it fires

  • agent 59%
  • story 22%
  • python 14%
  • chat 4%
  • worksheet 1%

Makes the model say

listing_letter I let will:

Makes it not say

In TheyirreturnisWhat checks around

Strongest examples

On text the dictionary never trained on. The underlined token is where it fires most.

  1. agent 3.53
    'double_al' is not defined. Did you mean: 'double_
  2. agent 3.35
    upper_ase' is not defined. Did you mean: 'upper_
  3. agent 3.21
    'lenggth' is not defined. Did you mean: 'length'?
  4. agent 3.15
    'totaal' is not defined. Did you mean: 'total'?
  5. python 3.0
    (nums) / len(nums)⏎```⏎⏎>>> mean_list([2,
  6. agent 2.93
    Did you mean: 'negate'?
  7. agent 2.84
    def mean_list(nums):
  8. agent 2.82
    <|user_start|>Create stats.py with two functions: mean_list that returns the
  9. python 2.81
    ⏎ROBOT⏎```<|assistant_end|><|bos|>Bug: `average_list` should return
  10. python 2.81
    OBOT⏎78⏎<|bos|>```python⏎def mean_list(nums):
  11. python 2.8
    (nums) / len(nums)⏎```⏎⏎>>> mean_list([5,
  12. python 2.79
    nums) / len(nums)`.⏎⏎```python⏎def mean_list(nums):