INDEX

Explanations

conjunctions that link ideas or phrases together

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_L9_Hcat_z_lr1.20e-03_l11.20e+00_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.9.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

0:0.08

1:0.03

2:0.04

3:0.11

4:0.04

5:0.14

6:0.05

7:0.10

8:0.21

9:0.03

10:0.08

11:0.03

Negative Logits

 Cyber

-2.41

tackle

-2.40

 Euros

-2.29

NSA

-2.27

ð

-2.21

 Identity

-2.20

 Fedora

-2.18

 SNAP

-2.17

videos

-2.16

icum

-2.15

POSITIVE LOGITS

 Henderson

2.93

 Shinra

2.68

 Kerr

2.51

 Kahn

2.49

 strugg

2.40

 leng

2.39

NetMessage

2.38

 Clifford

2.31

 radial

2.24

 Tsuk

2.24

Activations Density 0.000%

conjunctions that link ideas or phrases together

No Comments

No Known Activations