INDEX

Explanations

contractions of "do not" or similar phrases expressing negative sentiment

oai_token-act-pair · gpt-3.5-turbo

negations in statements

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

artney

-0.78

Engineers

-0.69

ersed

-0.68

 Citiz

-0.67

 Parenthood

-0.67

DeliveryDate

-0.65

 Passage

-0.63

 Compass

-0.63

 pione

-0.62

atur

-0.61

POSITIVE LOGITS

 kidding

1.03

 necessarily

0.93

 ashamed

0.91

 myself

0.90

 sure

0.89

 gonna

0.87

 afraid

0.86

oooooooo

0.84

hin

0.83

 believe

0.81

Activations Density 0.083%

contractions of "do not" or similar phrases expressing negative sentiment

negations in statements

No Comments

No Known Activations