INDEX

Explanations

phrases related to political events, specifically those involving the White House

oai_token-act-pair · gpt-3.5-turbo

references to the White House

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

onal

-0.65

ests

-0.63

PsyNetMessage

-0.61

 Euph

-0.61

 Horus

-0.61

 Vaughan

-0.60

ones

-0.59

 Siberian

-0.59

 Merlin

-0.58

module

-0.58

POSITIVE LOGITS

wife

0.90

House

0.87

 House

0.85

holder

0.84

hower

0.83

wives

0.81

 Office

0.80

 Intelligence

0.79

hyde

0.78

 Judiciary

0.76

Activations Density 0.016%

phrases related to political events, specifically those involving the White House

references to the White House

No Comments

No Known Activations