INDEX

Explanations

the word "typical" along with accompanying descriptions or attributes

oai_token-act-pair · gpt-3.5-turbo

references to common or average scenarios

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Racing

-0.66

iffe

-0.65

 Publishing

-0.63

raine

-0.63

otte

-0.62

AI

-0.60

Leaks

-0.60

faith

-0.59

ener

-0.58

oner

-0.58

POSITIVE LOGITS

 typical

3.53

 Typical

2.16

typ

2.03

 stereotypical

1.88

 usual

1.81

 characteristic

1.55

 normal

1.51

 average

1.47

 traditional

1.45

 customary

1.44

Activations Density 0.013%

the word "typical" along with accompanying descriptions or attributes

references to common or average scenarios

No Comments

No Known Activations