INDEX

Explanations

instances of the word "of"

oai_token-act-pair · gpt-3.5-turbo

phrases that refer to different "versions of" something

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

ibling

-0.82

resy

-0.79

oji

-0.76

 teasp

-0.76

urers

-0.75

ktop

-0.72

rences

-0.71

ering

-0.69

entimes

-0.67

erity

-0.66

POSITIVE LOGITS

 course

0.73

 history

0.62

thood

0.62

reality

0.59

 Meier

0.59

 Horton

0.58

 Chicken

0.58

 theirs

0.57

 Christianity

0.57

 Divinity

0.56

Activations Density 0.105%

instances of the word "of"

phrases that refer to different "versions of" something

No Comments

No Known Activations