INDEX

Explanations

very specific words, especially searching for long, specific combinations of characters

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "word" and its variations 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 @ 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

 Ryder

-0.75

ocr

-0.75

ocal

-0.69

icably

-0.66

ramid

-0.65

inav

-0.64

alia

-0.64

arij

-0.63

asio

-0.63

icas

-0.63

POSITIVE LOGITS

word

1.15

words

1.15

ptr

0.99

ãĥ£

0.85

DragonMagazine

0.85

acters

0.83

sworth

0.83

terday

0.81

press

0.79

mask

0.79

Activations Density 0.004%

very specific words, especially searching for long, specific combinations of characters

occurrences of the word "word" and its variations in the text

No Comments

No Known Activations