INDEX

Explanations

written messages or signs

oai_token-act-pair · gpt-3.5-turbo

references to notes and signs

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

å§«

-0.86

 Cosponsors

-0.70

ivities

-0.67

å£«

-0.67

 Flavoring

-0.66

kus

-0.66

imester

-0.66

 passions

-0.65

ventures

-0.64

sites

-0.63

POSITIVE LOGITS

 stating

1.63

 saying

1.54

 explaining

1.49

 telling

1.42

 informing

1.38

 indicating

1.36

 proclaiming

1.34

 advising

1.33

 declaring

1.31

 describing

1.30

Activations Density 0.312%

written messages or signs

references to notes and signs

No Comments

No Known Activations