INDEX

Explanations

proper nouns related to specific individuals

oai_token-act-pair · gpt-3.5-turbo

names and references related to notable individuals

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

last names that follow a title or a first name

oai_token-act-pair · gpt-4-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 buckle

-0.78

 Parenthood

-0.68

 gearing

-0.62

 Scandinavian

-0.60

 Bulgarian

-0.60

 recess

-0.59

 Universities

-0.59

 Valve

-0.58

 Civilization

-0.58

 Babylon

-0.58

POSITIVE LOGITS

gerald

1.18

 Fitz

1.03

avorite

0.95

allery

0.90

ezvous

0.87

patrick

0.85

schild

0.85

enegger

0.85

sonian

0.85

roy

0.84

Activations Density 0.015%

proper nouns related to specific individuals

names and references related to notable individuals

last names that follow a title or a first name

No Comments

No Known Activations