INDEX

Explanations

references to explosive devices and violent incidents

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

explosions, bombs, grenades, mines, attacks

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

Chham

-0.42

 übersch

-0.39

moveToNext

-0.39

Theo

-0.38

Jordan

-0.38

そこは

-0.38

 sogget

-0.37

 Jordan

-0.37

 نسبت

-0.36

 bilang

-0.36

POSITIVE LOGITS

 grenades

0.49

 explosives

0.46

 BorderRadius

0.46

MessageTagHelper

0.45

aarrggbb

0.43

SequentialGroup

0.41

 grenade

0.40

 bombings

0.40

 circulaire

0.40

 bombing

0.40

Activations Density 0.006%

references to explosive devices and violent incidents

explosions, bombs, grenades, mines, attacks

No Comments

No Known Activations