INDEX

Explanations

phrases related to user-generated content and personal engagement

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

instructions followed by requests

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

DockStyle

-0.35

juries

-0.33

tableLayoutPanel

-0.33

 endeavor

-0.33

 Lake

-0.32

 inner

-0.32

beginPath

-0.31

ass

-0.31

 ấn

-0.31

SSA

-0.31

POSITIVE LOGITS

 Infórmanos

0.77

 Савезне

0.71

angliski

0.66

 Biôgrafia

0.61

 للاسماء

0.56

 Keuangan

0.56

majánló

0.56

KommentareTeilen

0.56

 سكانية

0.55

bootstrapcdn

0.54

Activations Density 0.015%

phrases related to user-generated content and personal engagement

instructions followed by requests

No Comments

No Known Activations