INDEX

Explanations

phrases associated with beliefs or opinions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "believes."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

practice

-0.68

pmwiki

-0.68

 Pavilion

-0.67

insert

-0.67

med

-0.67

cloth

-0.66

ast

-0.66

nec

-0.64

effect

-0.63

cycl

-0.62

POSITIVE LOGITS

olate

0.79

 believes

0.78

ieve

0.74

ieves

0.73

ieved

0.70

aram

0.69

 believe

0.69

rase

0.69

oglu

0.67

uristic

0.67

Activations Density 0.012%

phrases associated with beliefs or opinions

instances of the word "believes."

No Comments

No Known Activations