INDEX

Explanations

instances of the word "spoof" or related words

oai_token-act-pair · gpt-3.5-turbo

terms related to humorous or comedic elements

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

å£«

-0.77

 revolutions

-0.73

 fullest

-0.72

KT

-0.71

 Occupations

-0.70

INAL

-0.70

 Cells

-0.68

ARC

-0.66

apsed

-0.66

IVER

-0.65

POSITIVE LOGITS

oof

1.53

estern

0.95

terness

0.94

arser

0.90

entimes

0.90

ness

0.86

ahs

0.85

eful

0.84

ertodd

0.82

edly

0.81

Activations Density 0.006%

instances of the word "spoof" or related words

terms related to humorous or comedic elements

No Comments

No Known Activations