INDEX

Explanations

referenced individuals and groups in a specific context related to events or actions

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

preceding "which" or "who"

np_acts-logits-general · gemini-2.0-flash

who have, which was, a complete

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-16k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

 ویکی‌پدی

-0.66

الحياه

-0.62

oa̍t

-0.56

 saites

-0.55

"}")

-0.55

 Biôgrafia

-0.52

 ویکی‌آمباردا

-0.52

 ffilmiau

-0.52



-0.52

ésultats

-0.51

POSITIVE LOGITS

beten

0.38

ConstraintMaker

0.33

новништво

0.33

 single

0.32

rungsseite

0.30

AddTagHelper

0.30

Button

0.30

Kilder

0.30

intervento

0.30

put

0.30

Activations Density 0.499%

referenced individuals and groups in a specific context related to events or actions

preceding "which" or "who"

who have, which was, a complete

No Comments

No Known Activations