INDEX

Explanations

words related to safety and security measures or concepts, especially regarding potential risks or threats

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

regulatory terms and concepts related to safety measures

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

1:0.03

2:0.29

3:0.06

4:0.18

5:0.05

6:0.04

7:0.03

8:0.05

9:0.09

10:0.06

11:0.03

Negative Logits

エル

-1.29

 turb

-1.18

pheus

-1.16

 Struggle

-1.16

 Ukrain

-1.15

chrom

-1.14

��

-1.13

 Trin

-1.13

 ILCS

-1.13

��

-1.12

POSITIVE LOGITS

iland

1.45

Secure

1.44

WARE

1.40

allas

1.27

PLIED

1.26

Safety

1.26

anding

1.25

illance

1.21

ournal

1.20

Catalog

1.19

Activations Density 0.002%

words related to safety and security measures or concepts, especially regarding potential risks or threats

regulatory terms and concepts related to safety measures

No Comments

No Known Activations