INDEX

Explanations

phrases indicating intentions or meanings behind actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 7-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_7/width_16k/average_l0_69

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.7.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

 EconPapers

-0.80

ьаж

-0.70

 censiti

-0.69

 sepsis

-0.64

 LCCN

-0.63

ViewFeatures

-0.62

orescence

-0.61

contentLoaded

-0.61

personalised

-0.61

Risks

-0.60

POSITIVE LOGITS

 intended

1.15

 meant

1.06

 designed

0.99

meant

0.97

intended

0.97

 предназна

0.93

 bedoeld

0.91

 destinés

0.89

 purpose

0.89

Intended

0.83

Activations Density 0.265%

phrases indicating intentions or meanings behind actions

No Comments

No Known Activations