INDEX

Explanations

information in longer texts, mentioning various individuals, places, and activities with a focus on academic, legal, and technical contexts

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

phrases indicating positions or titles of academia or authority

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-att-kk

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.04

1:0.02

2:0.18

3:0.05

4:0.29

5:0.04

6:0.02

7:0.02

8:0.10

9:0.11

10:0.04

11:0.02

Negative Logits

 recapt

-1.41

 curtains

-1.38

ptoms

-1.33

oğ

-1.31

 dots

-1.29

uctions

-1.28

 rentals

-1.27

��

-1.25

anqu

-1.25

 roundup

-1.24

POSITIVE LOGITS

cius

1.47

Weather

1.47

staff

1.32

hematically

1.32

 IEEE

1.29

Marx

1.24

 Gutenberg

1.24

 humanities

1.23

 Degree

1.22

student

1.22

Activations Density 0.019%

information in longer texts, mentioning various individuals, places, and activities with a focus on academic, legal, and technical contexts

phrases indicating positions or titles of academia or authority

No Comments

No Known Activations