INDEX

Explanations

phrases indicating superlatives or rankings, particularly related to quality or performance

oai_token-act-pair · gpt-3.5-turbo

phrases indicating ranking or being among the best in various categories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

_-

-0.73

atari

-0.68

Items

-0.67

Query

-0.66

oult

-0.64

rod

-0.64

raid

-0.64

izoph

-0.62

ãĥĩ

-0.60

ALSE

-0.59

POSITIVE LOGITS

 existence

1.22

 Europe

1.14

 history

1.12

 terms

1.08

 America

1.08

 town

1.05

 world

1.03

 Africa

0.90

 humankind

0.90

 Asia

0.89

Activations Density 0.072%

phrases indicating superlatives or rankings, particularly related to quality or performance

phrases indicating ranking or being among the best in various categories

No Comments

No Known Activations