INDEX

Explanations

references to attacks, particularly in a violent context

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

"attack" or "attacks"

np_acts-logits-general · gemini-2.0-flash

attack on

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 zelve

-0.54

 Picchu

-0.50

sidemargin

-0.47

 kautta

-0.47

jalá

-0.45

ruimte

-0.44

 laiko

-0.44

 guías

-0.44

 acreditar

-0.43

 ricev

-0.42

POSITIVE LOGITS

 attack

1.25

attack

1.14

 Attack

1.09

 attacks

1.05

 ATTACK

0.99

Attack

0.98

 attacking

0.96

ATTACK

0.95

attacks

0.93

 attacked

0.92

Activations Density 0.210%

references to attacks, particularly in a violent context

"attack" or "attacks"

attack on

No Comments

No Known Activations