INDEX

Explanations

phrases indicating something being topped with specific ingredients

oai_token-act-pair · gpt-3.5-turbo

the occurrence of the word "topped" in various contexts, particularly relating to food and rankings

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Trials

-0.72

Nou

-0.72

 Loan

-0.67

Ì

-0.67

otive

-0.66

 Infinite

-0.66

å£«

-0.66

 Duty

-0.65

QR

-0.65

Import

-0.65

POSITIVE LOGITS

 topped

1.28

 topping

1.25

mast

0.92

TOP

0.88

ackets

0.85

 tops

0.85

xual

0.82

nesday

0.80

xtap

0.78

 notch

0.78

Activations Density 0.008%

phrases indicating something being topped with specific ingredients

the occurrence of the word "topped" in various contexts, particularly relating to food and rankings

No Comments

No Known Activations