INDEX

Explanations

acronyms or specific keywords

oai_token-act-pair · gpt-3.5-turbo

the name of a specific brand or company

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

yip

-0.77

udeb

-0.71

maid

-0.67

 beware

-0.67

 punitive

-0.66

quartered

-0.66

ename

-0.66

aukee

-0.65

isconsin

-0.65

license

-0.64

POSITIVE LOGITS

0.97

 crow

0.85

 Rubin

0.79

 Ches

0.69

 Kraft

0.63

<+

0.63

 Hein

0.62

 Ceres

0.62

ATL

0.62

 Mets

0.61

Activations Density 0.000%

acronyms or specific keywords

the name of a specific brand or company

No Comments

No Known Activations