INDEX

Explanations

mentions of the word "mill" or derivatives like "millennial" and "million."

oai_token-act-pair · gpt-3.5-turbo

references to "millions" and variations of the term

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

lde

-0.76

 tailored

-0.72

 fitting

-0.70

 sting

-0.69

 vain

-0.69

 tang

-0.68

 selective

-0.66

 slippery

-0.66

 segregated

-0.65

 compelling

-0.65

POSITIVE LOGITS

enium

1.15

Mill

1.14

mill

1.10

ennial

1.07

enn

1.05

isec

1.02

 ILCS

0.98

imet

0.95

ENN

0.95

oton

0.93

Activations Density 0.006%

mentions of the word "mill" or derivatives like "millennial" and "million."

references to "millions" and variations of the term

No Comments

No Known Activations