INDEX

Explanations

various forms and contexts of the word "trick."

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

trick

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 9-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_9/width_131k/average_l0_121

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.9.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

 Marsden

-0.39

 yaşam

-0.39

Â

-0.38

Â

-0.37

 Weiden

-0.37

 Spalding

-0.37

 Hayden

-0.37

orene

-0.36

 Conserv

-0.36

 Bous

-0.36

POSITIVE LOGITS

 trick

2.44

 Trick

2.23

trick

2.16

Trick

2.13

 tricks

1.88

 Tricks

1.75

tricks

1.67

 truco

1.42

 tricked

1.33

 trucos

1.30

Activations Density 0.006%

various forms and contexts of the word "trick."

trick

No Comments

No Known Activations