INDEX

Explanations

text related to user interactions involving clicking actions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "click" indicating user interactions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Chancellor

-0.71

utive

-0.68

-0.67

von

-0.64

 Wein

-0.63

 Luxem

-0.63

ASC

-0.62

 Abram

-0.62

Zam

-0.60

Wol

-0.60

POSITIVE LOGITS

click

1.18

 clicks

0.95

 click

0.95

alore

0.91

lish

0.86

hent

0.84

 clicked

0.83

Click

0.80

urated

0.78

 clicking

0.76

Activations Density 0.011%

text related to user interactions involving clicking actions

instances of the word "click" indicating user interactions

No Comments

No Known Activations