INDEX

Explanations

words and phrases related to jokes and humorous content

oai_token-act-pair · gpt-3.5-turbo

Triggered by @bot

mentions and contexts related to jokes

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-att-kk

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.3.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.02

1:0.03

2:0.06

3:0.06

4:0.02

5:0.07

6:0.09

7:0.32

8:0.04

9:0.04

10:0.09

11:0.10

Negative Logits

onga

-1.37

undle

-1.27

arantine

-1.26

FTWARE

-1.21

bis

-1.19

idth

-1.19

igent

-1.18

imura

-1.18

hold

-1.16

 refurb

-1.16

POSITIVE LOGITS

 trivia

1.59

 humor

1.47

 jokes

1.45

 satir

1.39

 comedy

1.39

 humorous

1.35

 satire

1.33

 comedic

1.29

 parody

1.26

 Stupid

1.25

Activations Density 0.040%

words and phrases related to jokes and humorous content

mentions and contexts related to jokes

No Known Activations