INDEX

Explanations

references to geographical directions, specifically focusing on the concept of "north."

oai_token-act-pair · gpt-3.5-turbo

references to geographical directions, particularly "north"

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

inventoryQuantity

-0.87

TABLE

-0.87

TYPE

-0.83

PRESS

-0.80

ulous

-0.79

aminer

-0.78

argon

-0.74

ellation

-0.74

tons

-0.74

 CLSID

-0.72

POSITIVE LOGITS

ward

1.27

west

1.00

 hemisphere

0.99

 west

0.98

side

0.96

 east

0.96

wards

0.95

bound

0.93

shore

0.92

urst

0.92

Activations Density 0.021%

references to geographical directions, specifically focusing on the concept of "north."

references to geographical directions, particularly "north"

No Comments

No Known Activations