INDEX

Explanations

negative descriptions or criticisms

oai_token-act-pair · gpt-3.5-turbo

terms that express lack of value or effectiveness

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

Swe

-0.73

eston

-0.71

anwhile

-0.70

opsis

-0.69

ramid

-0.67

avia

-0.67

uph

-0.66

 Carol

-0.65

arthy

-0.64

 Brewer

-0.63

POSITIVE LOGITS

 useless

1.14

 idiots

0.92

nesses

0.86

 waste

0.85

NESS

0.83

 pointless

0.83

 worthless

0.81

 meaningless

0.80

seless

0.80

 superflu

0.80

Activations Density 0.007%

negative descriptions or criticisms

terms that express lack of value or effectiveness

No Comments

No Known Activations