INDEX

Explanations

questions directed at the user, especially those beginning with "do you" or "what do you."

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

answer markers

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

makeConstraints

-0.36

elemField

-0.36

IntoConstraints

-0.36

 useAppContext

-0.35

 Италијани

-0.34

 EconPapers

-0.32

mäß

-0.32

 étendue

-0.32

出版年

-0.32

قایناقلار

-0.30

POSITIVE LOGITS

ьаж

0.60

EndProject

0.51

#+#

0.50

 ActivatedRoute

0.47

GeoNames

0.46

 snippetHide

0.45

 pháp

0.44

┹

0.44

SizeF

0.44

 confidential

0.44

Activations Density 0.003%

questions directed at the user, especially those beginning with "do you" or "what do you."

answer markers

No Comments

No Known Activations