INDEX

Explanations

references to scientific journals, specifically the journal "Science"

oai_token-act-pair · gpt-3.5-turbo

mentions of the journal "Science."

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

arding

-0.75

onymous

-0.73

ivalent

-0.71

 timers

-0.67

ardless

-0.65

ieft

-0.64

upon

-0.64

teen

-0.63

 convincing

-0.62

arted

-0.62

POSITIVE LOGITS

 Fiction

1.30

 fiction

0.99

 Journalism

0.86

 Science

0.83

uality

0.81

 Medicine

0.77

craft

0.77

 Labs

0.76

 Talent

0.75

Pwr

0.75

Activations Density 0.015%

references to scientific journals, specifically the journal "Science"

mentions of the journal "Science."

No Comments

No Known Activations