INDEX

Explanations

proper nouns, specifically the name "Nelson"

oai_token-act-pair · gpt-3.5-turbo

references to the name "Nelson."

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

fare

-0.85

Â§

-0.77

BOOK

-0.75

Transfer

-0.73

orical

-0.71

¾

-0.71

¿

-0.71

ãĤ¼

-0.71

FML

-0.71

ASED

-0.69

POSITIVE LOGITS

 Mandela

1.54

 Nelson

1.36

 Rockefeller

0.84

otine

0.83

illo

0.79

 reefs

0.76

Sau

0.75

elson

0.74

sey

0.72

inson

0.71

Activations Density 0.005%

proper nouns, specifically the name "Nelson"

references to the name "Nelson."

No Comments

No Known Activations