INDEX

Explanations

words related to being surprised or amazed

oai_token-act-pair · gpt-3.5-turbo

instances of the word "behold" and related phrases indicating attention or observation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

•shakespeare

surprise

•layer 3 - surprise

No Comments

Negative Logits

 MacArthur

-0.70

 Hiroshima

-0.70

 Scotia

-0.66

CDC

-0.64

 Atkinson

-0.63

 McCann

-0.63

 Hurricanes

-0.63

 medic

-0.62

 Diesel

-0.62

 Kaiser

-0.61

POSITIVE LOGITS

 behold

1.33

entimes

1.12

ers

0.97

ardless

0.92

!--

0.89

htaking

0.88

masters

0.88

ering

0.84

enance

0.84

owship

0.83

Activations Density 0.008%

words related to being surprised or amazed

instances of the word "behold" and related phrases indicating attention or observation

No Comments

No Known Activations