INDEX

Explanations

actions related to protection or manipulation

oai_token-act-pair · gpt-3.5-turbo

phrases related to protection, manipulation, and influence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

osponsors

-0.63

eele

-0.62

 guiActiveUnfocused

-0.62

interstitial

-0.61

ursday

-0.61

onew

-0.59

GV

-0.59

coon

-0.56

cano

-0.56

angler

-0.55

POSITIVE LOGITS

 everything

1.03

 every

1.03

 EVERY

0.98

and

0.97

 innumerable

0.95

 themselves

0.93

 countless

0.91

 them

0.89

 everyone

0.88

 everybody

0.88

Activations Density 0.461%

actions related to protection or manipulation

phrases related to protection, manipulation, and influence

No Comments

No Known Activations