INDEX

Explanations

repeated references to "we" and "he" in the context of religious or moral assertions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 1-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_1/width_16k/average_l0_102

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.1.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

Moq

-0.74

 Adair

-0.66

Egl

-0.65

 campagnes

-0.63

 polvere

-0.61

 cumplido

-0.59

ONESIA

-0.59

 themſelves

-0.59

 sqlSession

-0.58

 foncé

-0.58

POSITIVE LOGITS

We

1.08

He

0.78

Hochspringen

0.75

WE

0.70

She

0.63

MemoryWarning

0.63

EqualsAnd

0.59

 They

0.59

 }}"></

0.57

ühner

0.55

Activations Density 0.163%

repeated references to "we" and "he" in the context of religious or moral assertions

No Comments

No Known Activations