INDEX

Explanations

mentions of legal actions involving various individuals

oai_token-act-pair · gpt-3.5-turbo

instances of the word "and."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ridges

-0.68

mus

-0.67

Contents

-0.66

mos

-0.62

izen

-0.62

ital

-0.60

reality

-0.58

imens

-0.58

agos

-0.58

irin

-0.58

POSITIVE LOGITS

 Associates

1.04

 Sons

0.90

 others

0.88

his

0.79

ERSON

0.78

 associates

0.73

 consequently

0.73

 other

0.72

 thereby

0.71

 pals

0.70

Activations Density 0.241%

mentions of legal actions involving various individuals

instances of the word "and."

No Comments

No Known Activations