INDEX

Explanations

acronyms related to intelligence or investigative organizations

oai_token-act-pair · gpt-3.5-turbo

references to intelligence agencies or associated abbreviations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Hiroshima

-0.74

 Cthulhu

-0.72

ize

-0.68

selves

-0.67

ãĥī

-0.67

days

-0.67

wards

-0.66

perature

-0.66

 Swedish

-0.66

 contempor

-0.66

POSITIVE LOGITS

IB

1.07

ANY

1.04

ANE

1.00

EW

0.99

ERC

0.98

ORN

0.97

terness

0.95

ARB

0.94

ERN

0.93

JJ

0.93

Activations Density 0.004%

acronyms related to intelligence or investigative organizations

references to intelligence agencies or associated abbreviations

No Comments

No Known Activations