INDEX

Explanations

mentions of investigating or looking into dishonesty and corruption

oai_token-act-pair · gpt-3.5-turbo

questions and mentions of specific social or political issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

xual

-0.75

pta

-0.72

mounted

-0.67

urch

-0.64

ents

-0.63

 disabled

-0.63

 normally

-0.61

fur

-0.61

una

-0.61

otally

-0.60

POSITIVE LOGITS

Detailed

0.85

 VIDEOS

0.76

Politics

0.73

Nap

0.73

Advertisements

0.72

Latest

0.72

NPR

0.71

Nov

0.71

Quotes

0.69

Discover

0.67

Activations Density 0.210%

mentions of investigating or looking into dishonesty and corruption

questions and mentions of specific social or political issues

No Comments

No Known Activations