INDEX

Explanations

words related to humiliation or embarrassment

oai_token-act-pair · gpt-3.5-turbo

references to specific geographical locations or names

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

IR

-0.75

NRS

-0.67

 Privacy

-0.64

Ã®

-0.62

LIC

-0.62

 summons

-0.62

MU

-0.60

 drunk

-0.60

 Pluto

-0.60

 refund

-0.60

POSITIVE LOGITS

aden

4.12

oat

2.25

 Aden

1.97

atten

1.88

 Deng

0.93

hemat

0.93

aten

0.92

acet

0.90

 Admir

0.85

aina

0.84

Activations Density 0.030%

words related to humiliation or embarrassment

references to specific geographical locations or names

No Comments

No Known Activations