INDEX

Explanations

references to strong interests or fixations on specific topics or activities

oai_token-act-pair · gpt-3.5-turbo

references to obsession and strong interests

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

stood

-0.84

 Kavanaugh

-0.72

Reviewed

-0.67

reens

-0.65

Medic

-0.63

arnaev

-0.62

received

-0.60

DragonMagazine

-0.59

casts

-0.58

ģ«

-0.58

POSITIVE LOGITS

ishly

0.84

ilyn

0.81

 fascination

0.81

obs

0.78

 fascinated

0.74

 fixation

0.74

uously

0.74

 obsession

0.73

 obsessed

0.72

ively

0.72

Activations Density 0.039%

references to strong interests or fixations on specific topics or activities

references to obsession and strong interests

No Comments

No Known Activations