INDEX

Explanations

buildings that house specific entities or activities

oai_token-act-pair · gpt-3.5-turbo

references to "houses" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

nesota

-0.71

PsyNetMessage

-0.68

 Tone

-0.66

TSA

-0.66

xon

-0.65

 clipboard

-0.65

READ

-0.64

ean

-0.64

iasis

-0.63

iferation

-0.62

POSITIVE LOGITS

chool

1.31

 houses

1.19

houses

1.04

wives

0.95

 dwellings

0.93

boats

0.91

 estates

0.88

poons

0.86

hops

0.86

 buildings

0.85

Activations Density 0.005%

buildings that house specific entities or activities

references to "houses" in various contexts

No Comments

No Known Activations