INDEX

Explanations

words related to rankings or levels of achievement

oai_token-act-pair · gpt-3.5-turbo

instances of the letter "r" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Lobby

-0.71

 Bulldogs

-0.65

creen

-0.65

hower

-0.64

 Sons

-0.64

 Nieto

-0.63

 cardio

-0.62

 compe

-0.61

 confines

-0.59

 EntityItem

-0.59

POSITIVE LOGITS

umbling

1.27

angers

1.25

ascal

1.24

ipper

1.21

haps

1.21

ipples

1.20

attle

1.20

umbled

1.19

ifling

1.19

umbles

1.18

Activations Density 0.027%

words related to rankings or levels of achievement

instances of the letter "r" in various contexts

No Comments

No Known Activations