INDEX

Explanations

references to the term "Hum" at varying intensities

oai_token-act-pair · gpt-3.5-turbo

references to humor or humorous content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

cape

-0.68

OHN

-0.67

 EntityItem

-0.67

hips

-0.66

 Peaks

-0.65

 approve

-0.64

 DARK

-0.64

 ---------

-0.64

ECTION

-0.62

holders

-0.61

POSITIVE LOGITS

pty

1.09

mers

1.00

iliation

0.99

ming

0.98

ankind

0.98

hum

0.97

orously

0.96

ility

0.94

Hum

0.91

ulus

0.90

Activations Density 0.009%

references to the term "Hum" at varying intensities

references to humor or humorous content

No Comments

No Known Activations