INDEX

Explanations

references to hoaxes and pranks in the text

oai_token-act-pair · gpt-3.5-turbo

references to hoaxes and pranks

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

oyal

-0.73

avid

-0.71

definition

-0.69

arching

-0.69

 obligated

-0.66

uv

-0.66

arity

-0.66

ibr

-0.66

affer

-0.65

foreseen

-0.65

POSITIVE LOGITS

 hoax

1.27

 spoof

0.91

gee

0.88

ulence

0.85

 prank

0.84

sters

0.83

mong

0.79

ãĥ¼ãĥ«

0.78

gey

0.76

etooth

0.74

Activations Density 0.012%

references to hoaxes and pranks in the text

references to hoaxes and pranks

No Comments

No Known Activations