INDEX

Explanations

mentions of specific names, possibly related to politics or law

oai_token-act-pair · gpt-3.5-turbo

mentions of specific characters and individuals associated with a certain narrative or context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

printed

-0.77

acl

-0.76

geries

-0.70

formed

-0.69

information

-0.68

ocrine

-0.67

ribute

-0.66

cycles

-0.65

hedral

-0.63

ASE

-0.62

POSITIVE LOGITS

 Goodman

1.23

 Saul

0.93

 Berman

0.86

Ö

0.79

 Katz

0.77

iflower

0.77

éĹĺ

0.76

ÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤ

0.74

 Misc

0.74

TING

0.74

Activations Density 0.013%

mentions of specific names, possibly related to politics or law

mentions of specific characters and individuals associated with a certain narrative or context

No Comments

No Known Activations