INDEX

Explanations

phrases indicating the end or conclusion of something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "end."

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

Not in Any Lists

No Comments

Negative Logits

ppo

-0.84

chy

-0.67

issance

-0.66

CHA

-0.66

kers

-0.65

 Byrd

-0.63

kson

-0.61

IGHTS

-0.61

kaya

-0.60

aution

-0.60

POSITIVE LOGITS

angered

1.07

lich

1.01

angering

0.99

owment

0.99

orph

0.95

ocrine

0.92

ocrin

0.90

urance

0.88

end

0.86

orse

0.83

Activations Density 0.025%

phrases indicating the end or conclusion of something

instances of the word "end."

No Comments

No Known Activations