INDEX

Explanations

individuals in the news or public figures

oai_token-act-pair · gpt-3.5-turbo

references to various professions and public figures

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

nih

-0.69

ecause

-0.66

mbuds

-0.63

 Helpful

-0.61

 Parables

-0.60

ysics

-0.59

noon

-0.59

idding

-0.58

 Liter

-0.58

si

-0.57

POSITIVE LOGITS

 underwent

1.00

 admits

0.94

 extraord

0.91

 endured

0.91

 insists

0.88

 believes

0.86

 opted

0.86

 joked

0.85

 prefers

0.85

 stumbled

0.83

Activations Density 0.143%

individuals in the news or public figures

references to various professions and public figures

No Comments

No Known Activations