INDEX

Explanations

instances where something is allowed or enabled

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate ability or permission

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

Sax

-0.72

 Leviathan

-0.69

 Notting

-0.67

ravis

-0.67

ĸļ

-0.64

é¾įå¥ĳå£«

-0.63

lass

-0.63

lr

-0.60

 Springer

-0.59

mma

-0.59

POSITIVE LOGITS

SpaceEngineers

0.80

undai

0.66

tera

0.66

otiation

0.63

SEE

0.62

 wearer

0.61

 slightest

0.61

 flexibility

0.61

TEXTURE

0.60

 withdrawals

0.60

Activations Density 0.349%

instances where something is allowed or enabled

phrases that indicate ability or permission

No Comments

No Known Activations