INDEX

Explanations

terms related to publications and various social topics

oai_token-act-pair · gpt-3.5-turbo

references to various forms of media and culture

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

 antioxid

-0.53

ptoms

-0.50

ptive

-0.49

IDA

-0.47

alore

-0.46

 cryst

-0.46

 pestic

-0.46

 Lumpur

-0.45

ILA

-0.45

 COURT

-0.44

POSITIVE LOGITS

ictionary

0.52

ĨĴ

0.51

ween

0.51

 Barbie

0.50

enment

0.50

¥µ

0.47

glers

0.47

Nazis

0.47

 stereotype

0.46

 Parables

0.45

Activations Density 6.867%

terms related to publications and various social topics

references to various forms of media and culture

No Comments

No Known Activations