INDEX

Explanations

names of individuals or figures in newsworthy contexts

oai_token-act-pair · gpt-3.5-turbo

mentions of specific individuals and terms related to weapons or operative activities

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

 Pegasus

-0.73

 Gors

-0.72

abase

-0.69

 Azerb

-0.65

 Emblem

-0.65

 Judaism

-0.65

gypt

-0.64

 Jonah

-0.64

 looph

-0.63

 Torah

-0.63

POSITIVE LOGITS

ãĤ¼ãĤ¦ãĤ¹

0.77

ickson

0.70

INO

0.70

ordes

0.68

noticed

0.67

iard

0.65

software

0.65

 chim

0.64

å¾

0.64

interstitial

0.63

Activations Density 0.001%

names of individuals or figures in newsworthy contexts

mentions of specific individuals and terms related to weapons or operative activities

No Comments

No Known Activations