INDEX

Explanations

instances of the letter "B" with high activation values

oai_token-act-pair · gpt-3.5-turbo

references to the letter "B"

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Gutenberg

-0.85

oldemort

-0.78

terday

-0.75

 Lerner

-0.69

ãĥĨãĤ£

-0.66

rolet

-0.66

velt

-0.65

Whereas

-0.63

Supported

-0.62

 Enlightenment

-0.62

POSITIVE LOGITS

rug

1.02

razen

0.96

erk

0.95

aked

0.94

urt

0.92

uch

0.92

rief

0.92

rien

0.91

BA

0.91

uh

0.90

Activations Density 0.035%

instances of the letter "B" with high activation values

references to the letter "B"

No Comments

No Known Activations