INDEX

Explanations

references to a particular person named Hogan

oai_token-act-pair · gpt-3.5-turbo

specific names, particularly 'Hogan' and 'Gawker'

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

yrim

-0.75

aeda

-0.73

reen

-0.71

ulet

-0.68

nesota

-0.68

aution

-0.67

iterranean

-0.67

olulu

-0.66

ql

-0.65

PsyNetMessage

-0.65

POSITIVE LOGITS

 Hogan

1.53

azard

0.76

 Hulk

0.73

lich

0.72

AMA

0.71

 Feldman

0.71

buster

0.70

lord

0.69

 McMahon

0.69

enegger

0.69

Activations Density 0.007%

references to a particular person named Hogan

specific names, particularly 'Hogan' and 'Gawker'

No Comments

No Known Activations