INDEX

Explanations

references to survey questions and responses

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

open-ended questions and prompts

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

Negative Logits

VersionUID

-0.37

StatusOK

-0.34

 Einfluß

-0.33

 initComponents

-0.33

 CreateTagHelper

-0.33

intenant

-0.33

 campanha

-0.32

setChecked

-0.31

 Gleichzeitig

-0.31

 Belastung

-0.31

POSITIVE LOGITS

extAlignment

0.68

ClCompile

0.55

argout

0.55

 CURIAM

0.53

 okuyayım

0.48

 responses

0.48

Responses

0.47

ResponseBody

0.46

 Anſ

0.46

 scrittura

0.46

Activations Density 0.015%

references to survey questions and responses

open-ended questions and prompts

No Known Activations