INDEX

Explanations

terms related to things within a specific country

oai_token-act-pair · gpt-3.5-turbo

references to domestic-related topics or issues

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

uyomi

-0.89

uden

-0.82

*/(

-0.75

hra

-0.73

osen

-0.70

 Kinnikuman

-0.68

edin

-0.68

veyard

-0.67

zzo

-0.64

ums

-0.64

POSITIVE LOGITS

 Violence

0.96

 domestic

0.93

 Domestic

0.88

 affairs

0.86

violence

0.80

 violence

0.80

ãĤª

0.79

ãĤ¢ãĥ«

0.76

estic

0.76

 abuser

0.72

Activations Density 0.010%

terms related to things within a specific country

references to domestic-related topics or issues

No Comments

No Known Activations