INDEX

Explanations

terms related to user interaction, specifically clicking

oai_token-act-pair · gpt-3.5-turbo

words related to clicking and interaction with links or buttons

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

 Chancellor

-0.74

utive

-0.73

 Yard

-0.67

apolis

-0.62

é¾įå

-0.62

Wol

-0.61

 Variety

-0.61

 Warsaw

-0.60

 Croat

-0.60

 Aden

-0.59

POSITIVE LOGITS

lish

0.95

prints

0.85

antry

0.85

dress

0.77

oola

0.77

esome

0.76

elaide

0.75

erk

0.75

jriwal

0.75

ularity

0.74

Activations Density 0.016%

terms related to user interaction, specifically clicking

words related to clicking and interaction with links or buttons

No Comments

No Known Activations