INDEX

Explanations

references to historical figures and titles

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

claims table figures

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

expandindo

-0.73

SharedDtor

-0.57

 EconPapers

-0.54

 ujednoznacz

-0.54

 Numerade

-0.52

 дописавши

-0.51

ыгана

-0.50

RTCK

-0.49

Географиясе

-0.49

webElementGuid

-0.49

POSITIVE LOGITS

mbggenerated

0.55

 Connell

0.37

Cześć

0.36

IRQn

0.36

 Wikiseite

0.34

 уж

0.34

WillAppear

0.33

Aholisi

0.33

 ayudarte

0.32

Activations Density 0.146%

references to historical figures and titles

claims table figures

No Comments

No Known Activations