INDEX

Explanations

phrases that express contradictions or highlight hypocrisy in political discourse

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

but, but would be better

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

 Gegens

-0.54

 spowod

-0.48

 rotina

-0.45

Reprodução

-0.45

Rujuakan

-0.45

oficina

-0.44

llavero

-0.43

 hırka

-0.43

的一声

-0.43

 devido

-0.42

POSITIVE LOGITS

MLLoader

0.50

 StatelessWidget

0.46

 calendriers

0.45

jfree

0.44

Източници

0.43

Бележки

0.42

 gyhoeddwyd

0.42

 שוליים

0.41

nodoc

0.41

BufferException

0.40

Activations Density 0.101%

phrases that express contradictions or highlight hypocrisy in political discourse

but, but would be better

No Comments

No Known Activations