INDEX

Explanations

the word "pit" followed by a high activation number

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of "pit bull."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.71

 Carbuncle

-0.69

 Lauder

-0.66

 challeng

-0.65

 proport

-0.64

IGH

-0.63

 Feinstein

-0.62

lihood

-0.61

 Polo

-0.60

 Leilan

-0.59

POSITIVE LOGITS

iful

1.35

cair

1.28

ifully

1.27

iless

1.24

cher

1.08

bull

0.94

uit

0.90

iable

0.88

adium

0.87

reatment

0.86

Activations Density 0.039%

the word "pit" followed by a high activation number

repeated mentions of "pit bull."

No Comments

No Known Activations