INDEX

Explanations

specific numerical references or citation markers in academic contexts

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L10_Hcat_z_lr1.20e-03_l11.30e+00_ds24576_bs4096_dc1.00e-05_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.10.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.04

1:0.03

2:0.05

3:0.23

4:0.10

5:0.12

6:0.03

7:0.04

8:0.05

9:0.12

10:0.10

11:0.03

Negative Logits

 toget

-2.62

.''

-2.42

theless

-2.38

uana

-2.29

manship

-2.21

erie

-2.19

 internationally

-2.19

icut

-2.19

 bour

-2.16

 later

-2.16

POSITIVE LOGITS

Pt

2.64

 Proc

2.60

 dstg

2.54

akin

2.49

Pi

2.38

Sac

2.32

Inv

2.31

 Tuls

2.27

Sac

2.23

Chapter

2.15

Activations Density 0.002%

specific numerical references or citation markers in academic contexts

No Known Activations