INDEX

Explanations

phrases related to actions taken against illegal activities or measures put in place to protect specific groups or entities

oai_token-act-pair · gpt-3.5-turbo

references to "sanctuary" or related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

llor

-0.81

 Nebula

-0.74

drivers

-0.68

ï¸ı

-0.67

ives

-0.65

drive

-0.65

 destro

-0.65

REE

-0.64

llers

-0.64

urned

-0.63

POSITIVE LOGITS

ctuary

1.46

ction

1.10

ctions

0.98

San

0.97

itary

0.94

gha

0.92

itized

0.88

 Francisco

0.88

terior

0.87

anton

0.87

Activations Density 0.018%

phrases related to actions taken against illegal activities or measures put in place to protect specific groups or entities

references to "sanctuary" or related concepts

No Comments

No Known Activations