INDEX

Explanations

phrases related to inappropriate behavior and consequences

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

OLOGY

-0.73

 Trojan

-0.68

 Waterloo

-0.67

 Roose

-0.66

enegger

-0.66

eways

-0.63

WM

-0.63

esan

-0.62

 Printing

-0.61

Kev

-0.61

POSITIVE LOGITS

inent

0.92

 abst

0.90

ain

0.90

ainer

0.87

inance

0.86

ention

0.85

rences

0.85

atory

0.83

ained

0.82

aining

0.81

Activations Density 0.027%

phrases related to inappropriate behavior and consequences

No Comments

No Known Activations