INDEX

Explanations

There is a focus on identifying comedic and satirical elements in the text

oai_token-act-pair · gpt-3.5-turbo

references to parody and satire

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

å£«

-1.09

elsen

-0.83

 Borders

-0.81

erto

-0.79

 cryst

-0.79

 membranes

-0.78

 carriers

-0.77

 opio

-0.77

igree

-0.76

 Folder

-0.73

POSITIVE LOGITS

 spoof

1.09

 satir

1.08

 banter

1.07

 parody

1.05

 humor

1.05

 mocking

1.00

 humorous

0.99

 joking

0.99

 humour

0.98

 prank

0.97

Activations Density 0.074%

There is a focus on identifying comedic and satirical elements in the text

references to parody and satire

No Known Activations