INDEX

Explanations

terms related to titles of products or projects

oai_token-act-pair · gpt-3.5-turbo

references to games or media titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ajor

-0.71

gm

-0.70

 Sabha

-0.70

 intestine

-0.65

addafi

-0.64

agan

-0.64

ths

-0.62

hod

-0.61

 spice

-0.60

ucl

-0.60

POSITIVE LOGITS

manship

1.15

 titles

1.00

ãĥĩ

0.96

itles

0.86

paces

0.85

marks

0.81

¥µ

0.81

pad

0.80

title

0.78

tions

0.78

Activations Density 0.015%

terms related to titles of products or projects

references to games or media titles

No Comments

No Known Activations