INDEX

Explanations

problem

The neuron consistently activates on occurrences of “problem” (and related terms indicating a difficulty), i.e. when the text frames something as a problem or issue.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

न

1.66

ان

1.46

1.32

ك

1.25

۹

1.18

ন

1.17

 exige

1.16

 оцен

1.13

 وزار

1.11

۹

1.11

POSITIVE LOGITS

ग्रस्त

1.30

 PROBLEMS

1.30

Problems

1.13

PROBLE

1.13

 problems

1.12

 Problems

1.09

MIB

1.08

́t

1.06

 问题

1.05

্ভব

1.03

Activations Density 0.044%