INDEX

Explanations

phrases related to politics and governmental positions

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

MpServer

-0.74

lot

-0.71

Î³

-0.66

ratom

-0.66

lighting

-0.65

essen

-0.65

endish

-0.62

Medium

-0.61

doors

-0.60

 informants

-0.59

POSITIVE LOGITS

 Barack

1.10

clinton

1.07

ially

1.06

 Obama

0.95

 Lyndon

0.92

ial

0.91

 Donald

0.85

IAL

0.83

 Jinping

0.82

 hopeful

0.81

Activations Density 1.045%

phrases related to politics and governmental positions

No Comments

No Known Activations