INDEX

Explanations

adjectives indicating quality or importance

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

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

fromnode

-0.53

 jScrollPane

-0.49

 confetti

-0.48

▻

-0.47

 alguém

-0.47

 pharmacists

-0.47

▍

-0.46

daß

-0.46

 bikers

-0.44

 pharmacies

-0.44

POSITIVE LOGITS

تقاوى

0.56

الدراسه

0.47

Hochspringen

0.46

<bos>

0.44

alogical

0.43

webElementXpaths

0.42

 EconPapers

0.41

hoedd

0.40

 strongly

0.40

andExpect

0.39

Activations Density 0.078%

adjectives indicating quality or importance

No Comments

No Known Activations