INDEX

Explanations

references to doors

oai_token-act-pair · gpt-3.5-turbo

references to doors

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Hots

-0.66

abi

-0.65

azeera

-0.65

ting

-0.65

ciating

-0.65

umbnail

-0.63

ortium

-0.62

ontent

-0.62

thood

-0.61

ency

-0.60

POSITIVE LOGITS

door

1.24

bell

1.15

steps

1.06

holes

1.04

 doors

1.00

 door

0.99

hole

0.95

 opener

0.95

 Door

0.95

doors

0.91

Activations Density 0.012%

references to doors

references to doors

No Comments

No Known Activations