INDEX

Explanations

words related to security or protection, particularly with a focus on the concepts of safety and safeguarding

oai_token-act-pair · gpt-3.5-turbo

forms of the suffixes 'ured' and 'uring', indicating a focus on actions or states of being related to security or condition

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ingham

-0.92

soDeliveryDate

-0.80

ITNESS

-0.76

owler

-0.69

udeb

-0.65

strings

-0.65

endium

-0.64

ADS

-0.61

Parser

-0.60

ever

-0.58

POSITIVE LOGITS

ures

0.75

 Brach

0.75

ure

0.73

tsky

0.72

uring

0.68

anship

0.67

ulic

0.66

ually

0.65

URE

0.65

ured

0.65

Activations Density 0.020%

words related to security or protection, particularly with a focus on the concepts of safety and safeguarding

forms of the suffixes 'ured' and 'uring', indicating a focus on actions or states of being related to security or condition

No Comments

No Known Activations