INDEX

Explanations

terms related to harmful and toxic substances or bacteria

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

harmful, toxic, pathogenic

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

Ref

-0.40

venez

-0.38

gü

-0.37

longValue

-0.36

மே

-0.36

cain

-0.36

 Wind

-0.36

 where

-0.35

REF

-0.35

 Chair

-0.35

POSITIVE LOGITS

Datuak

0.59

exitRule

0.54

kloped

0.54

DOCTYPE

0.54

 الرياضيه

0.53

úgó

0.52

 poisonous

0.51

有害

0.51

 parís

0.51

 harmful

0.50

Activations Density 0.049%

terms related to harmful and toxic substances or bacteria

harmful, toxic, pathogenic

No Comments

No Known Activations