INDEX

Explanations

proper nouns, such as names of people, places, and organizations

oai_token-act-pair · gpt-3.5-turbo

phrases that confirm information or reports

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

BuyableInstoreAndOnline

-0.79

 "$:/

-0.72

 ILCS

-0.67

ĸļ

-0.64

 Shutterstock

-0.62

 undone

-0.61

interstitial

-0.61

lest

-0.61

 Glover

-0.60

 Canaver

-0.59

POSITIVE LOGITS

 existence

1.06

 validity

0.98

 authenticity

0.97

 presence

0.93

 paternity

0.86

 involvement

0.82

 legitimacy

0.81

 identity

0.81

 affiliation

0.78

 importance

0.76

Activations Density 0.222%

proper nouns, such as names of people, places, and organizations

phrases that confirm information or reports

No Comments

No Known Activations