INDEX

Explanations

words related to censorship or inappropriate content

oai_token-act-pair · gpt-3.5-turbo

words related to the concept of "uncertainty" or "unknown."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

hyde

-0.71

 desk

-0.68

tery

-0.64

++++++++++++++++

-0.63

 Monthly

-0.61

MENT

-0.61

 upholding

-0.60

 menstrual

-0.58

 Solitaire

-0.58

Ò

-0.58

POSITIVE LOGITS

redited

1.45

ategor

1.43

orrect

1.38

ritical

1.37

ooked

1.30

ount

1.29

outh

1.28

apped

1.27

ivil

1.25

ustom

1.25

Activations Density 0.023%

words related to censorship or inappropriate content

words related to the concept of "uncertainty" or "unknown."

No Comments

No Known Activations