INDEX

Explanations

financial and numerical information, such as rankings, quantities, or values

oai_token-act-pair · gpt-3.5-turbo

lists of ranked items or statistics presented with accompanying numerical values

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

izen

-0.70

»Ĵ

-0.70

oÄŁ

-0.68

erent

-0.67

enegger

-0.67

avez

-0.67

obyl

-0.66

nih

-0.66

skirts

-0.66

crop

-0.64

POSITIVE LOGITS

 surpass

1.32

 trailing

1.20

 averaging

1.19

behind

1.19

 boasting

1.16

 surpassed

1.16

 topping

1.04

 beating

1.03

 ranking

1.02

 according

1.00

Activations Density 0.200%

financial and numerical information, such as rankings, quantities, or values

lists of ranked items or statistics presented with accompanying numerical values

No Comments

No Known Activations