INDEX

Explanations

proper nouns, particularly names of individuals and notable entities

New Auto-Interp

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

Plots

Explanation

Show Test Field

Default Test Text

IFrame

Link

Not in Any Lists

Head Attr Weights

0:0.03

1:0.10

2:0.02

3:0.02

4:0.03

5:0.40

6:0.02

7:0.01

8:0.04

9:0.14

10:0.10

11:0.03

Negative Logits

PN

-1.70

 endemic

-1.59

 metic

-1.56

 Trop

-1.50

Balt

-1.43

 unexpl

-1.43

apons

-1.43

xp

-1.41

ppo

-1.40

ngth

-1.38

POSITIVE LOGITS

1.76

etts

1.71

ulously

1.69

and

1.55

\":

1.53

 photographed

1.49

icipated

1.48

±

1.47

iott

1.44

ihara

1.42

Activations Density 0.078%

proper nouns, particularly names of individuals and notable entities

No Activations Uploaded