INDEX

Explanations

content related to a series with sequential parts

oai_token-act-pair · gpt-3.5-turbo

phrases indicating parts of a series or articles

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

animate

-0.70

 receipt

-0.67

 CLSID

-0.66

 symbol

-0.65

 wink

-0.64

enance

-0.63

 fingerprints

-0.63

ibles

-0.62

æ©

-0.62

 consulate

-0.61

POSITIVE LOGITS

 installment

1.04

 installments

1.03

isode

0.88

 series

0.87

series

0.86

Interview

0.84

 roundup

0.81

 debunk

0.79

 Discussion

0.78

PDATED

0.78

Activations Density 0.297%

content related to a series with sequential parts

phrases indicating parts of a series or articles

No Comments

No Known Activations