INDEX

Explanations

highlights or important points within text

oai_token-act-pair · gpt-3.5-turbo

phrases and terms related to highlights or key events

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

essee

-0.92

eri

-0.91

interstitial

-0.86

ice

-0.81

ews

-0.74

oking

-0.73

nery

-0.73

merga

-0.73

aut

-0.72

esthes

-0.72

POSITIVE LOGITS

 Highlights

1.22

Fi

0.83

 Battery

0.80

ailable

0.74

 Improvement

0.71

 SOFTWARE

0.69

 Wilmington

0.69

 Includes

0.69

 Significant

0.68

 Phant

0.68

Activations Density 0.016%

highlights or important points within text

phrases and terms related to highlights or key events

No Comments

No Known Activations