INDEX

Explanations

capitalized abbreviations followed by a number

oai_token-act-pair · gpt-3.5-turbo

brand names and specific software or technologies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

mble

-0.76

hement

-0.76

lihood

-0.71

 âĢº

-0.63

ocrin

-0.61

çĶ°

-0.61

=#

-0.60

assetsadobe

-0.59

 Tradable

-0.59

ãĥ©ãĥ³

-0.59

POSITIVE LOGITS

 Citadel

0.58

 Dome

0.58

ahime

0.58

hovah

0.55

ogenesis

0.54

 hubs

0.53

topia

0.53

Gs

0.53

ModLoader

0.52

Hai

0.52

Activations Density 0.506%

capitalized abbreviations followed by a number

brand names and specific software or technologies

No Comments

No Known Activations