INDEX

Explanations

HTML tags and formatting indicators

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

HTML paragraph and end-of-turn markers

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

Jereo

-0.47

 Italijani

-0.33

avras

-0.33

 Bünd

-0.32

 innego

-0.31

ésult

-0.31

toHaveBeenCalled

-0.31

apunov

-0.30

 lesión

-0.30

DatabaseError

-0.30

POSITIVE LOGITS

findpost

0.59

 EconPapers

0.57

TagMode

0.57

rungsseite

0.56

 <<<<<<<<<<<<<<

0.55

 transférez

0.55

 defaultstate

0.55

MessageOf

0.55

 BrowserModule

0.54

ResumeLayout

0.52

Activations Density 0.002%

HTML tags and formatting indicators

HTML paragraph and end-of-turn markers

No Comments

No Known Activations