INDEX

Explanations

instances where the text conveys surprise or the potential for surprise

oai_token-act-pair · gpt-3.5-turbo

instances of the word "surprised"

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

IFrame

Link

•layer 3 - surprise

No Comments

Negative Logits

©¶æ

-0.81

ciplinary

-0.78

amins

-0.74

ngth

-0.72

tein

-0.71

uria

-0.71

atra

-0.71

ignty

-0.70

vertisement

-0.70

hid

-0.70

POSITIVE LOGITS

 Squid

0.82

 onlook

0.74

terson

0.72

 Surprise

0.72

NESS

0.71

imaru

0.69

 silence

0.69

 surprised

0.67

 enough

0.66

Jol

0.65

Activations Density 0.014%

instances where the text conveys surprise or the potential for surprise

instances of the word "surprised"

No Comments

No Known Activations