INDEX

Explanations

mentions of specific locations or addresses

oai_token-act-pair · gpt-3.5-turbo

articles and demonstratives referring to nouns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Vaugh

-0.71

 Seym

-0.70

 helicop

-0.69

ilaterally

-0.66

 destro

-0.65

anamo

-0.62

 Moroc

-0.59

onto

-0.59

ãĥ¼ãĥĨ

-0.58

 vulner

-0.57

POSITIVE LOGITS

 largeDownload

0.84

 nutshell

0.76

 Rock

0.58

Ĥİ

0.58

 heels

0.56

 Pixel

0.53

 grate

0.52

 Rapids

0.50

 peninsula

0.50

 Baltimore

0.50

Activations Density 0.211%

mentions of specific locations or addresses

articles and demonstratives referring to nouns

No Comments

No Known Activations