INDEX

Explanations

occurrences of the word "Edison" or the abbreviation "ORN" followed by a sequence of numbers

oai_token-act-pair · gpt-3.5-turbo

references to specific names, particularly associated with notable figures or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

CB

-0.80

 Pupp

-0.70

Tags

-0.70

DOM

-0.68

TF

-0.67

 Monteneg

-0.67

deals

-0.66

FLAG

-0.64

TY

-0.64

TT

-0.64

POSITIVE LOGITS

 Edison

4.14

ORN

1.25

 Maxwell

1.23

umers

1.08

 Lowell

0.97

ioxide

0.97

auri

0.96

 Lauder

0.94

 transformer

0.86

 silicon

0.85

Activations Density 0.049%

occurrences of the word "Edison" or the abbreviation "ORN" followed by a sequence of numbers

references to specific names, particularly associated with notable figures or entities

No Comments

No Known Activations