INDEX

Explanations

personalized messages indicating something specific is meant for the reader

oai_token-act-pair · gpt-3.5-turbo

references to the second person, particularly the word "you" in various contexts

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

ĸļ

-0.82

ipal

-0.80

entimes

-0.77

aughed

-0.73

ice

-0.67

apo

-0.66

ariat

-0.63

pite

-0.62

urtle

-0.62

oche

-0.61

POSITIVE LOGITS

 guys

1.44

tub

1.28

RS

1.15

're

1.08

Tube

0.89

NG

0.86

 yourselves

0.83

sir

0.81

 yourself

0.81

hei

0.79

Activations Density 0.109%

personalized messages indicating something specific is meant for the reader

references to the second person, particularly the word "you" in various contexts

No Comments

No Known Activations