INDEX

Explanations

sequences of underscores or spaces used for formatting or indicating empty content

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

code comments and declarations

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

 للمعارف

-0.68

évaluateur

-0.59

 للاسماء

-0.57

 كومونز

-0.56

ectady

-0.50

setChecked

-0.48

MLLoader

-0.48



-0.47

المشاركات

-0.47

__(/*!

-0.46

POSITIVE LOGITS

SourceChecksum

0.44

yntaxException

0.42

modelBuilder

0.42

toHaveBeenCalled

0.38

 Wicidata

0.37

 محفوظة

0.36

StructEnd

0.36

 conduct

0.36

 hansen

0.36

 Conduct

0.35

Activations Density 0.000%

sequences of underscores or spaces used for formatting or indicating empty content

code comments and declarations

No Comments

No Known Activations