INDEX

Explanations

phrases related to inventions or discoveries

oai_token-act-pair · gpt-3.5-turbo

instances of the word "invented."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

Thom

-0.75

sen

-0.67

isite

-0.63

HUD

-0.63

ander

-0.62

union

-0.60

rooms

-0.59

alon

-0.58

nee

-0.58

bourne

-0.58

POSITIVE LOGITS

 invented

1.12

æ©

0.96

 concoct

0.90

ĸļ

0.90

ãĥ¼ãĥĨãĤ£

0.89

ulence

0.87

glers

0.79

 invent

0.76

 Doodle

0.76

uristic

0.75

Activations Density 0.014%

phrases related to inventions or discoveries

instances of the word "invented."

No Comments

No Known Activations