INDEX

Explanations

mentions of technology and data-related activities, especially in the context of surveillance and information gathering

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

the word "in" as it appears in various contexts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.04

1:0.02

2:0.11

3:0.06

4:0.27

5:0.07

6:0.03

7:0.02

8:0.14

9:0.09

10:0.05

11:0.02

Negative Logits

rush

-1.33

heed

-1.23

behind

-1.21

 shame

-1.17

adra

-1.16

 bombed

-1.14

 boarded

-1.12

bley

-1.11

lease

-1.09

vious

-1.09

POSITIVE LOGITS

cellaneous

1.40

ournal

1.39

UCT

1.38

CLASSIFIED

1.37

================================================================

1.30

ESSION

1.30

erning

1.29

Offline

1.29

 culinary

1.28

ormal

1.27

Activations Density 0.002%

mentions of technology and data-related activities, especially in the context of surveillance and information gathering

the word "in" as it appears in various contexts

No Comments

No Known Activations