INDEX

Explanations

references to a specific location, "Mal"

oai_token-act-pair · gpt-3.5-turbo

references to the term "Mal"

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

arettes

-0.81

ICLE

-0.76

soDeliveryDate

-0.73

OLOGY

-0.71

ACTED

-0.68

ellation

-0.68

RSS

-0.65

ä»£

-0.65

ptions

-0.64

OPLE

-0.64

POSITIVE LOGITS

icious

0.92

colm

0.89

imir

0.88

mal

0.86

inda

0.85

ibu

0.84

ignant

0.84

practice

0.83

mort

0.80

aques

0.79

Activations Density 0.006%

references to a specific location, "Mal"

references to the term "Mal"

No Comments

No Known Activations