INDEX

Explanations

words or phrases that denote events or significant occurrences

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

Head Attr Weights

0:0.02

1:0.01

2:0.23

3:0.05

4:0.05

5:0.03

6:0.33

7:0.03

8:0.03

9:0.05

10:0.07

11:0.03

Negative Logits

 causation

-1.49

natureconservancy

-1.45

ecause

-1.39

 haircut

-1.33

 indic

-1.32

alties

-1.29

 outp

-1.27

asking

-1.27

rity

-1.26

stown

-1.25

POSITIVE LOGITS

ennes

1.56

Dispatch

1.54

 Monaco

1.42

BuyableInstoreAndOnline

1.40

;;;;;;;;;;;;

1.40

 Alma

1.39

Stack

1.39

��

1.37

 Wilkinson

1.35

デ

1.33

Activations Density 0.002%

words or phrases that denote events or significant occurrences

No Known Activations