INDEX

Explanations

phrases indicating purpose or intention, often linked to the word "order."

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

purpose followed by action

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

Erreferentziak

-0.44

OG

-0.39

Carriera

-0.38

 Dodo

-0.37

ogie

-0.37

)";

-0.36

crimin

-0.35

logo

-0.34

 Rother

-0.34

rim

-0.34

POSITIVE LOGITS

Cyfarwyddwr

0.54

 vectorielle

0.53

ImageContext

0.51

 alimentaria

0.50

 tacón

0.50

TestBed

0.50

 curiosidad

0.49

 Italijanski

0.48

ImageField

0.48

 Waray

0.48

Activations Density 0.061%

phrases indicating purpose or intention, often linked to the word "order."

purpose followed by action

No Comments

No Known Activations