INDEX

Explanations

mentions of intense or extreme situations, often negative

oai_token-act-pair · gpt-3.5-turbo

quotations in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 overlook

-0.71

 favor

-0.66

 scheduled

-0.65

 shop

-0.65

 lookout

-0.64

 derby

-0.64

 disapprove

-0.64

 termin

-0.64

tro

-0.63

 adjud

-0.63

POSITIVE LOGITS

They

1.27

We

1.25

Our

1.15

It

1.13

1.12

There

1.12

Where

1.12

Sometimes

1.10

Too

1.10

Because

1.09

Activations Density 0.103%

mentions of intense or extreme situations, often negative

quotations in the text

No Comments

No Known Activations