INDEX

Explanations

references to publications and publishers

oai_token-act-pair · gpt-3.5-turbo

the phrase "published by" and related structures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ndum

-0.78

icides

-0.70

 manif

-0.70

quez

-0.69

aucuses

-0.68

osexual

-0.67

xual

-0.67

employment

-0.66

ptoms

-0.66

cffffcc

-0.66

POSITIVE LOGITS

 Warner

1.02

 Universal

0.99

 Univ

0.95

 Wizards

0.93

 Electronic

0.91

 STATS

0.90

 Columbia

0.90

 Carnegie

0.90

 Penguin

0.90

 Paramount

0.88

Activations Density 0.158%

references to publications and publishers

the phrase "published by" and related structures

No Comments

No Known Activations