INDEX

Explanations

Twitter usernames preceded by an "@" symbol

oai_token-act-pair · gpt-3.5-turbo

mentions of social media handles or usernames

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 orche

-0.71

agra

-0.70

 rust

-0.69

 elect

-0.69

 hust

-0.68

 immersion

-0.67

 methamphetamine

-0.67

 royalty

-0.63

bestos

-0.62

 Malk

-0.62

POSITIVE LOGITS

(@

1.20

#$

0.99

username

0.78

SourceFile

0.77

realDonaldTrump

0.77

sorry

0.75

deck

0.75

updated

0.75

NB

0.74

hello

0.74

Activations Density 0.019%

Twitter usernames preceded by an "@" symbol

mentions of social media handles or usernames

No Comments

No Known Activations