INDEX

Explanations

occurrences of comments or significant whitespace in code snippets

oai_token-act-pair · gpt-4o-mini Triggered by @bot

code structures and non-english characters

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

::_('

-0.64

MLLoader

-0.54

Hentet

-0.53

 Wikimedijinoj

-0.51

󠁣

-0.51

Diweddarwch

-0.50

➯

-0.49

runApp

-0.47

 createState

-0.45

⟬

-0.45

POSITIVE LOGITS

announce

0.37

abstractmethod

0.37

otherwise

0.36

íssima

0.36

ăsp

0.36

IsMutable

0.36

midrule

0.35

逅

0.35

 Tripp

0.35

VersionUID

0.35

Activations Density 0.004%

occurrences of comments or significant whitespace in code snippets

code structures and non-english characters

No Comments

No Known Activations