INDEX

Explanations

emotionally charged and evaluative words or phrases

oai_token-act-pair · gpt-3.5-turbo

descriptive phrases expressing strong opinions or emotions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

rongh

-0.82

execute

-0.78

idden

-0.78

obook

-0.76

orsi

-0.74

govtrack

-0.73

artney

-0.73

Downloadha

-0.73

bara

-0.72

foreseen

-0.71

POSITIVE LOGITS

huh

1.21

eh

0.95

tho

0.88

 coincidence

0.84

 kidding

0.83

 congr

0.81

 downside

0.80

ya

0.72

!!

0.72

 Kills

0.71

Activations Density 0.287%

emotionally charged and evaluative words or phrases

descriptive phrases expressing strong opinions or emotions

No Comments

No Known Activations