INDEX

Explanations

terms related to security and institutions

oai_token-act-pair · gpt-3.5-turbo

references to "identity" and "identities."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Beng

-0.59

mark

-0.58

chen

-0.57

 elbows

-0.57

 cass

-0.57

eng

-0.56

 brow

-0.55

back

-0.55

Nak

-0.55

Kop

-0.54

POSITIVE LOGITS

ITY

4.50

ITIES

3.51

ities

2.54

ity

2.38

itÃ©

1.96

ISM

1.57

idad

1.46

IFIED

1.44

IZE

1.44

NESS

1.31

Activations Density 0.012%

terms related to security and institutions

references to "identity" and "identities."

No Comments

No Known Activations