INDEX

Explanations

references to advertising and advertisers

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ologies

-0.77

 Patriarch

-0.72

VK

-0.72

APS

-0.72

arity

-0.71

stood

-0.69

 confinement

-0.65

ologically

-0.64

 surviving

-0.64

 turtle

-0.61

POSITIVE LOGITS

vertis

1.09

 advertisers

0.95

vertising

0.93

eering

0.91

elaide

0.87

 advertis

0.85

ounce

0.84

elson

0.81

 Advertising

0.75

ele

0.74

Activations Density 0.006%

references to advertising and advertisers

No Comments

No Known Activations