INDEX

Explanations

proper nouns or names belonging to people

oai_token-act-pair · gpt-3.5-turbo

repeated references to specific individuals and characters

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

element

-0.69

ws

-0.68

SS

-0.68

CSS

-0.67

 LGBTQ

-0.67

 housing

-0.66

ant

-0.65

 Social

-0.64

ssl

-0.63

Ge

-0.63

POSITIVE LOGITS

 McCoy

4.24

 Spock

1.73

McC

1.56

 Scully

1.48

 McCartney

1.46

 Decker

1.21

 McCabe

1.21

 Picard

1.21

Pry

1.14

 Pryor

1.11

Activations Density 0.066%

proper nouns or names belonging to people

repeated references to specific individuals and characters

No Comments

No Known Activations