INDEX

Explanations

phrases related to online actions like signing up for something, taking action, and privacy policies

oai_token-act-pair · gpt-3.5-turbo

calls to action related to privacy and engagement

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 cond

-0.67

etheless

-0.66

fr

-0.65

 brill

-0.65

 appar

-0.63

bri

-0.62

 Hann

-0.61

cel

-0.61

ilion

-0.61

tains

-0.60

POSITIVE LOGITS

Brow

0.75

Month

0.74

Drive

0.72

 Payments

0.72

Trade

0.70

Pay

0.70

Breaking

0.67

Subscribe

0.67

Newsletter

0.66

 Rights

0.65

Activations Density 0.051%

phrases related to online actions like signing up for something, taking action, and privacy policies

calls to action related to privacy and engagement

No Comments

No Known Activations