INDEX

Explanations

company names or initials

oai_token-act-pair · gpt-3.5-turbo

instances of the abbreviation "GL" and related terms in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

taining

-0.71

cling

-0.68

cles

-0.68

tale

-0.64

ctl

-0.63

 Hiroshima

-0.63

ÄŁ

-0.63

hyde

-0.63

rano

-0.62

stood

-0.62

POSITIVE LOGITS

ength

1.11

ENN

0.99

ibrary

0.96

ASS

0.89

ocal

0.86

ossip

0.86

OSS

0.85

ibr

0.85

ocaust

0.84

FW

0.83

Activations Density 0.006%

company names or initials

instances of the abbreviation "GL" and related terms in the text

No Comments

No Known Activations