INDEX

Explanations

phrases indicating uncertainty or hesitation

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.07

1:0.02

2:0.22

3:0.09

4:0.21

5:0.05

6:0.03

7:0.02

8:0.08

9:0.10

10:0.05

11:0.01

Negative Logits

SpaceEngineers

-1.33

CFR

-1.32

 Flavoring

-1.30

TF

-1.22

COMPLE

-1.20

 Farming

-1.18

 Masonic

-1.18

cig

-1.18

DSL

-1.18

 サーティワン

-1.18

POSITIVE LOGITS

atsu

1.42

idia

1.41

 suspic

1.41

appa

1.36

issa

1.34

uum

1.32

Quantity

1.30

aki

1.29

ihara

1.28

adata

1.28

Activations Density 0.005%

phrases indicating uncertainty or hesitation

No Known Activations