INDEX

Explanations

terms related to power, authority, and control in political contexts

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

 Daylight

-0.69

asca

-0.67

 Hawking

-0.67

ruary

-0.67

thal

-0.65

akeru

-0.64

neys

-0.64

enegger

-0.63

TOM

-0.62

uni

-0.61

POSITIVE LOGITS

 levers

1.04

 wielded

1.00

 exercised

0.99

 exerted

0.94

holder

0.93

holding

0.89

 vested

0.88

 controlled

0.87

 delegated

0.85

 controlling

0.81

Activations Density 5.525%

terms related to power, authority, and control in political contexts

No Comments

No Known Activations