INDEX

Explanations

punctuation and formatting elements, particularly around photo credits and publication details

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_L5_Hcat_z_lr1.20e-03_l11.00e+00_ds49152_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

float32

Hook Name

blocks.5.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

1:0.02

2:0.03

3:0.11

4:0.05

5:0.04

6:0.04

7:0.33

8:0.06

9:0.09

10:0.07

11:0.08

Negative Logits

pled

-1.71

artifacts

-1.65

etheless

-1.56

 incorpor

-1.55

wagon

-1.52

 registrations

-1.52

arthed

-1.50

 acad

-1.47

eday

-1.43

prints

-1.43

POSITIVE LOGITS

..........

1.65

 Courtesy

1.42

 COURT

1.34

 Himself

1.34

›

1.31

ˈ

1.29

utterstock

1.28

 Known

1.28

github

1.27

00200000

1.24

Activations Density 0.087%

punctuation and formatting elements, particularly around photo credits and publication details

No Comments

No Known Activations