INDEX

Explanations

references to violent and tragic events involving mass harm to individuals

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

nos

-0.78

ilk

-0.77

yssey

-0.74

ĪĴ

-0.72

tis

-0.72

thens

-0.72

shire

-0.71

thy

-0.70

uana

-0.70

BILITIES

-0.70

POSITIVE LOGITS

achusetts

1.42

achus

0.96

imil

0.89

 quantities

0.84

acre

0.83

 exodus

0.83

 transit

0.83

aging

0.83

 incarceration

0.82

imation

0.79

Activations Density 0.053%

references to violent and tragic events involving mass harm to individuals

No Comments

No Known Activations