INDEX

Explanations

puns or wordplay in text

oai_token-act-pair · gpt-3.5-turbo

instances of the word "pun" or variations thereof

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

;;;;;;;;;;;;

-0.71

 Empires

-0.68

 Covenant

-0.68

 Consent

-0.67

Ae

-0.65

 Close

-0.65

 Transparency

-0.64

 Breach

-0.63

 Commons

-0.61

itutional

-0.61

POSITIVE LOGITS

isher

1.34

ishers

1.30

ting

1.13

ters

1.13

ter

1.12

cheon

1.09

cher

1.07

ishment

1.03

tering

1.00

pun

0.96

Activations Density 0.035%

puns or wordplay in text

instances of the word "pun" or variations thereof

No Comments

No Known Activations