INDEX

Explanations

phrases related to preventing, discouraging, disapproving, or expressing negativity

oai_token-act-pair · gpt-3.5-turbo

terms related to negative sentiment or disapproval

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

htaking

-0.92

iom

-0.91

ophon

-0.90

mble

-0.89

isters

-0.89

ural

-0.88

isting

-0.87

eson

-0.86

icians

-0.84

uncture

-0.84

POSITIVE LOGITS

 discour

0.71

 discouraging

0.71

 discourage

0.70

 inhib

0.70

AGE

0.69

 spoil

0.69

 discouraged

0.68

ADS

0.68

shy

0.67

 undermin

0.67

Activations Density 0.018%

phrases related to preventing, discouraging, disapproving, or expressing negativity

terms related to negative sentiment or disapproval

No Comments

No Known Activations