INDEX

Explanations

phrases related to speaking out or expressing opinions

oai_token-act-pair · gpt-3.5-turbo

references to voices and expressions from various groups or individuals, particularly in discussions of representation and inequality

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

amily

-0.70

onut

-0.69

hib

-0.69

 Kear

-0.69

etheless

-0.67

sis

-0.67

addy

-0.65

keyes

-0.64

slaught

-0.62

athing

-0.62

POSITIVE LOGITS

 voices

1.03

 recorder

1.02

 voice

1.00

voice

0.98

 louder

0.85

 Voice

0.79

 melody

0.76

holder

0.76

 heard

0.74

 chorus

0.73

Activations Density 0.016%

phrases related to speaking out or expressing opinions

references to voices and expressions from various groups or individuals, particularly in discussions of representation and inequality

No Comments

No Known Activations