INDEX

Explanations

words related to controversial or sensitive topics, such as racism, terrorism, legal disputes, and political issues

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

ullah

-0.70

ipation

-0.65

eers

-0.63

icio

-0.61

itu

-0.60

ingham

-0.58

cape

-0.58

ificent

-0.58

pole

-0.58

eer

-0.57

POSITIVE LOGITS

verning

1.14

 overboard

0.95

ggle

0.92

lems

0.92

©¶æ

0.88

vt

0.88

 forth

0.87

 ballistic

0.81

aded

0.81

etz

0.81

Activations Density 0.104%

words related to controversial or sensitive topics, such as racism, terrorism, legal disputes, and political issues

No Comments

No Known Activations