INDEX

Explanations

negations and phrases indicating a lack of something or a contradiction

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

negation and absence

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

-0.57

 calendriers

-0.53

 InputDecoration

-0.48

InstrumentedTest

-0.43

 StatefulWidget

-0.41

îné

-0.39

Efq

-0.38

TPS

-0.38

@[+][

-0.38

 houſe

-0.37

POSITIVE LOGITS

+#+#

0.63

 الحره

0.53

 snippetHide

0.52

 Rptr

0.47

 szolg

0.47

OGND

0.45

URLException

0.44

ביוגרפיה

0.44

OwnProperty

0.44

 betrokken

0.42

Activations Density 0.114%

negations and phrases indicating a lack of something or a contradiction

negation and absence

No Comments

No Known Activations