INDEX

Explanations

the term "widespread" indicating prevalence or extensive coverage

oai_token-act-pair · gpt-3.5-turbo

instances of the word "widespread" and its variations related to general consensus or occurrence in society

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

=-=-

-0.82

men

-0.81

estamp

-0.80

izons

-0.76

eters

-0.76

thur

-0.74

¶æ

-0.70

asus

-0.69

pent

-0.69

wives

-0.68

POSITIVE LOGITS

 misconception

0.96

 adoption

0.95

 acceptance

0.90

 dissemination

0.87

wealth

0.85

 occurrence

0.84

 acclaim

0.83

 availability

0.82

 condemnation

0.82

 misunderstanding

0.80

Activations Density 0.043%

the term "widespread" indicating prevalence or extensive coverage

instances of the word "widespread" and its variations related to general consensus or occurrence in society

No Comments

No Known Activations