INDEX

Explanations

words and phrases associated with actions or movements

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

actions followed by direction

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

 Masyarakat

-0.39

 among

-0.33

 représent

-0.33

ÁND

-0.33

śnia

-0.33

ACIONAL

-0.32

capan

-0.31

 Budaya

-0.30

 Misalnya

-0.30

 popular

-0.29

POSITIVE LOGITS

ſelben

0.86

<unused41>

0.86

<unused14>

0.86

[@BOS@]

0.86

<unused8>

0.86

<unused42>

0.85

<unused43>

0.85

<unused28>

0.85

<unused17>

0.85

<unused21>

0.85

Activations Density 0.056%

words and phrases associated with actions or movements

actions followed by direction

No Comments

No Known Activations