INDEX

Explanations

the word "null" with high activation values

oai_token-act-pair · gpt-3.5-turbo

instances of the term "null" and its related context in various scenarios

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Downloadha

-0.80

BOOK

-0.76

 livest

-0.75

hens

-0.73

ilitating

-0.72

xual

-0.70

asio

-0.69

akov

-0.69

 millenn

-0.69

IFT

-0.68

POSITIVE LOGITS

ptr

1.10

ifying

1.01

ifies

0.99

ities

0.97

ified

0.92

ality

0.92

ify

0.90

ity

0.89

 null

0.83

ifier

0.82

Activations Density 0.013%

the word "null" with high activation values

instances of the term "null" and its related context in various scenarios

No Comments

No Known Activations