INDEX

Explanations

various forms of the word "de" as in degrade, depleting, and deportation

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.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

 Flavoring

-0.79

 Wage

-0.77

uyomi

-0.68

 Showdown

-0.67

ilial

-0.67

ĸļ

-0.65

 Forward

-0.65

 Teacher

-0.65

CPC

-0.63

 Topics

-0.62

POSITIVE LOGITS

arest

1.27

 Blasio

1.26

cember

1.22

 facto

1.21

activate

1.19

activated

1.14

vel

1.10

vious

1.09

jected

1.09

utsche

1.06

Activations Density 0.021%

various forms of the word "de" as in degrade, depleting, and deportation

No Comments

No Known Activations