INDEX

Explanations

proper nouns, including names of people and places

oai_token-act-pair · gpt-3.5-turbo

names of notable individuals and characters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 disadvantage

-0.69

 disadvantages

-0.69

 RELEASE

-0.66

 decreases

-0.65

 deduction

-0.64

 advantages

-0.63

 wildfires

-0.62

profits

-0.62

 affects

-0.61

 dwellings

-0.60

POSITIVE LOGITS

 himself

1.24

Jr

1.16

enegger

1.11

who

1.00

 Himself

0.99

 whom

0.93

 herself

0.92

Sr

0.88

hetti

0.87

aka

0.86

Activations Density 0.369%

proper nouns, including names of people and places

names of notable individuals and characters

No Comments

No Known Activations