INDEX

Explanations

phrases associated with vulnerability or potential harm to individuals

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

New Auto-Interp

Top Features by Cosine Similarity

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

Not in Any Lists

No Comments

Head Attr Weights

0:0.02

1:0.03

2:0.06

3:0.06

4:0.02

5:0.03

6:0.24

7:0.23

8:0.07

9:0.06

10:0.08

11:0.05

Negative Logits

 Qualcomm

-1.12

 comprises

-1.10

 shootout

-1.10

 commenter

-1.05

 gradient

-1.02

 Golem

-1.02

 difference

-1.01

 correlation

-1.00

 respectively

-1.00

 Difference

-0.99

POSITIVE LOGITS

 herself

1.67

exempt

1.30

idge

1.23

interrupted

1.18

ilaterally

1.17

iscover

1.16

PHOTOS

1.15

ffield

1.14

chlor

1.12

ritch

1.11

Activations Density 0.061%

phrases associated with vulnerability or potential harm to individuals

No Comments

No Known Activations