INDEX

Explanations

phrases indicating an excessive or negative sentiment towards a particular subject

oai_token-act-pair · gpt-3.5-turbo

phrases emphasizing excessive quantity or severity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

enance

-0.83

quo

-0.73

 emer

-0.66

itatively

-0.65

è¦ļéĨĴ

-0.63

ograph

-0.63

 defending

-0.63

Ips

-0.62

 à¨

-0.61

farious

-0.61

POSITIVE LOGITS

Too

1.11

BuyableInstoreAndOnline

0.82

Enough

0.81

utonium

0.78

Too

0.77

ifax

0.72

bad

0.71

username

0.69

len

0.67

 Klux

0.67

Activations Density 0.006%

phrases indicating an excessive or negative sentiment towards a particular subject

phrases emphasizing excessive quantity or severity

No Comments

No Known Activations