INDEX

Explanations

proper nouns or names, particularly related to people

oai_token-act-pair · gpt-3.5-turbo

repeated names or references to specific individuals in the text

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

Aad

-0.71

Bie

-0.69

 Maya

-0.68

thing

-0.68

FACE

-0.68

âĢ¢âĢ¢

-0.66

 Afgh

-0.65

 answ

-0.64

 bubble

-0.64

 stink

-0.63

POSITIVE LOGITS

 Berger

1.99

 Macron

1.88

alle

1.76

 Robertson

1.34

Rou

1.24

LC

1.21

olin

1.21

oton

1.20

 Rogue

1.09

rift

1.08

Activations Density 0.034%

proper nouns or names, particularly related to people

repeated names or references to specific individuals in the text

No Comments

No Known Activations