INDEX

Explanations

repetitive or recurring phrases and concepts within discussions

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

how to ask

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

 invitamos

-0.42

SharedCtor

-0.38

 AssemblyCulture

-0.37

ammans

-0.37

 voudrais

-0.37

 trouvera

-0.35

 domanda

-0.34

 muñeca

-0.34

 kvä

-0.33

 saurait

-0.32

POSITIVE LOGITS

脚注の使い方

0.68

Vidite

0.68

 ***!

0.65

IBOutlet

0.52

transQ

0.47

 kaarangay

0.47



0.47

 dAtA

0.46

gulls

0.46

dze

0.46

Activations Density 0.038%

repetitive or recurring phrases and concepts within discussions

how to ask

No Comments

No Known Activations