INDEX

Explanations

the repeated use of the word "over."

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

over

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

*/].

-0.43

Princip

-0.42

itects

-0.42

 tjen

-0.41

 Principals

-0.41

 Cactus

-0.41

Eins

-0.40

 MwSt

-0.39

Tanz

-0.38

 fibroblast

-0.38

POSITIVE LOGITS

 over

1.15

 Over

1.09

over

1.09

 OVER

1.04

Over

0.98

OVER

0.89

kover

0.85

 över

0.79

 über

0.78

tover

0.72

Activations Density 0.057%

the repeated use of the word "over."

over

No Comments

No Known Activations