INDEX

Explanations

dates in a specific format

oai_token-act-pair · gpt-3.5-turbo

dates mentioned within the document

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

ACTED

-0.76

milo

-0.76

ropolitan

-0.75

rower

-0.73

Reviewer

-0.70

 diaper

-0.70

ROR

-0.69

akings

-0.68

 diapers

-0.67

 thirds

-0.64

POSITIVE LOGITS

Jan

1.03

itors

1.03

Jan

1.01

uary

1.01

vier

0.97

Feb

0.95

itor

0.94

Feb

0.87

Nov

0.87

uine

0.81

Activations Density 0.007%

dates in a specific format

dates mentioned within the document

No Comments

No Known Activations