INDEX

Explanations

hidden elements or information

oai_token-act-pair · gpt-3.5-turbo

instances of the word "hidden"

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

rists

-0.71

OPA

-0.70

oslav

-0.68

onduct

-0.67

Hug

-0.66

aq

-0.65

 TAMADRA

-0.65

 Amend

-0.65

 Ambro

-0.63

 Faul

-0.63

POSITIVE LOGITS

 hidden

3.54

hidden

2.80

Hidden

2.42

 Hidden

2.32

 concealed

2.07

 hiding

1.83

 unseen

1.72

hid

1.72

 invisible

1.71

 secret

1.67

Activations Density 0.013%

hidden elements or information

instances of the word "hidden"

No Comments

No Known Activations