INDEX

Explanations

phrases related to being without something

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the absence of authorization or consent

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

oola

-0.88

ulty

-0.77

lish

-0.75

mun

-0.75

raq

-0.74

ocations

-0.74

des

-0.74

omsky

-0.71

late

-0.70

assies

-0.69

POSITIVE LOGITS

lessly

0.87

 knowing

0.84

 regard

0.80

 mentioning

0.79

 blinking

0.78

 compromising

0.76

 forgetting

0.72

 interruption

0.71

 sacrificing

0.70

 Borders

0.69

Activations Density 0.038%

phrases related to being without something

phrases indicating the absence of authorization or consent

No Comments

No Known Activations