INDEX

Explanations

references to the reader or audience

oai_token-act-pair · gpt-3.5-turbo

references to the reader in a document

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

eryl

-0.78

 Parenthood

-0.72

chwitz

-0.72

undai

-0.71

igham

-0.68

roxy

-0.67

 Spears

-0.66

ackle

-0.65

rals

-0.64

rity

-0.64

POSITIVE LOGITS

 reader

1.21

 Reader

0.99

Reader

0.98

verse

0.87

 readers

0.85

reader

0.84

 Readers

0.83

 APPLIC

0.77

itatively

0.77

OGR

0.77

Activations Density 0.009%

references to the reader or audience

references to the reader in a document

No Comments

No Known Activations