INDEX

Explanations

words related to negative experiences or controversies

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

)</

-0.76

 shores

-0.68

 Signs

-0.68

 Reloaded

-0.61

CHAT

-0.61

 Ruins

-0.61

ovember

-0.59

HOU

-0.59

anwhile

-0.59

src

-0.59

POSITIVE LOGITS

 imaginable

0.87

 guiActiveUnfocused

0.78

manship

0.73

 populism

0.66

ivalry

0.64

iness

0.63

 whatsoever

0.62

esthesia

0.62

yip

0.61

anship

0.61

Activations Density 0.219%

words related to negative experiences or controversies

No Comments

No Known Activations