INDEX

Explanations

the author name "Drozdov" written in different contexts

oai_token-act-pair · gpt-3.5-turbo

words associated with notable individuals or names

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Scots

-0.71

LESS

-0.69

uckland

-0.66

sworth

-0.65

SAY

-0.65

 footing

-0.64

CLR

-0.64

 Parliament

-0.64

METHOD

-0.63

owship

-0.62

POSITIVE LOGITS

ewski

1.18

sky

1.09

owicz

1.08

nik

1.04

owski

1.04

inski

1.02

ansky

0.98

ovsky

0.98

Ã¡n

0.95

anski

0.95

Activations Density 0.214%

the author name "Drozdov" written in different contexts

words associated with notable individuals or names

No Comments

No Known Activations