INDEX

Explanations

references to radio broadcasts and news content

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

radio broadcasting

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

ple

-0.31

 sala

-0.28

))}

-0.28

tre

-0.28

 high

-0.27

cn

-0.27

得了

-0.27

cai

-0.26

 skewed

-0.26

 locations

-0.26

POSITIVE LOGITS

 rádio

0.79

 broadcasting

0.78

Radio

0.76

 feroit

0.72

 broadcasts

0.72

radio

0.72

 Radio

0.72

⟬

0.71

 radio

0.70

ſelf

0.70

Activations Density 0.051%

references to radio broadcasts and news content

radio broadcasting

No Comments

No Known Activations