INDEX

Explanations

sequences of markup or code elements

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

court citations and code

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

KindOfClass

-0.59

FunctionFlags

-0.56

AutoField

-0.50

addWidget

-0.50

ModelAdmin

-0.49

RTCF

-0.48

 مشين

-0.47

めでとう

-0.47

Viited

-0.45

帖最后由

-0.44

POSITIVE LOGITS

⮕

0.41

0.36

الحياه

0.36

GEBURTSDATUM

0.35

endaten

0.35

ifikationer

0.34

 autorytatywna

0.34

Ap

0.33

nastics

0.33

 нео

0.33

Activations Density 0.017%

sequences of markup or code elements

court citations and code

No Comments

No Known Activations

sequences of markup or code elements

court citations and code

No Comments

No Known Activations