INDEX

Explanations

words related to technology and media, specifically television and consumption

oai_token-act-pair · gpt-3.5-turbo

terms related to television and mourning

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

Not in Any Lists

No Comments

Negative Logits

 Maid

-0.78

geist

-0.71

REDACTED

-0.68

hill

-0.68

 Mong

-0.66

LESS

-0.64

 Rover

-0.63

 Harriet

-0.63

 Trooper

-0.63

 weights

-0.63

POSITIVE LOGITS

isions

1.28

uous

0.96

igrated

0.94

cipled

0.93

anced

0.93

opsis

0.93

ision

0.92

isting

0.89

ancies

0.87

orce

0.86

Activations Density 0.016%

words related to technology and media, specifically television and consumption

terms related to television and mourning

No Comments

No Known Activations