INDEX

Explanations

No Explanations Found

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

0:0.07

1:0.06

2:0.10

3:0.09

4:0.08

5:0.07

6:0.09

7:0.07

8:0.08

9:0.08

10:0.08

11:0.09

Negative Logits

 plans

-1.64

utical

-1.56

 sponsors

-1.56

 Dino

-1.56

 Scout

-1.54

BuyableInstoreAndOnline

-1.53

 Shuttle

-1.51

-1.50

 domains

-1.49

WWF

-1.49

POSITIVE LOGITS

hedon

1.79

>>>

1.75

>)

1.72

EStreamFrame

1.57

bon

1.56

yth

1.54

illo

1.54

ifiable

1.51

poral

1.50

�

1.48

Activations Density 0.000%

No Known Activations

This feature has no known activations.