INDEX

Explanations

dates and events mentioned in a specific format

oai_token-act-pair · gpt-3.5-turbo

numerical values and measurements

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

 suspic

-0.61

DragonMagazine

-0.59

ratulations

-0.56

isode

-0.56

 unbeliev

-0.55

usually

-0.52

bett

-0.52

nesty

-0.51

typically

-0.51

essing

-0.51

POSITIVE LOGITS

 Leilan

0.63

 Provincial

0.61

 Marketable

0.57

 Galile

0.56

 pandemonium

0.54

REDACTED

0.51

 understatement

0.51

ymes

0.50

 discord

0.50

 Spac

0.50

Activations Density 1.725%

dates and events mentioned in a specific format

numerical values and measurements

No Comments

No Known Activations