INDEX

Explanations

proper nouns or names ending with the suffix "ared"

oai_token-act-pair · gpt-3.5-turbo

instances of the word "ared" or related variations within contexts emphasizing fear or danger

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

Mit

-0.68

BIP

-0.66

alter

-0.62

IB

-0.61

 Bleach

-0.58

 refurb

-0.58

UME

-0.56

UX

-0.55

 Wilkinson

-0.55

 rebuilding

-0.55

POSITIVE LOGITS

ared

1.06

ares

0.92

paren

0.92

mares

0.81

emo

0.80

ndra

0.80

ansas

0.75

erness

0.73

iffs

0.73

itors

0.72

Activations Density 0.006%

proper nouns or names ending with the suffix "ared"

instances of the word "ared" or related variations within contexts emphasizing fear or danger

No Comments

No Known Activations