INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L9_Hcat_z_lr1.20e-03_l11.20e+00_ds24576_bs4096_dc1.00e-06_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.9.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

0:0.02

1:0.03

2:0.02

3:0.03

4:0.02

5:0.40

6:0.02

7:0.06

8:0.03

9:0.16

10:0.01

11:0.14

Negative Logits

 裏�

-3.37

pse

-3.36

 thermal

-3.33

 rigging

-3.19

 resh

-3.15

 temperature

-2.97

 therm

-2.92

May

-2.68

RC

-2.66

Temperature

-2.58

POSITIVE LOGITS

ebook

3.19

Newsletter

2.90

Netflix

2.86

 newcomers

2.76

 Portable

2.73

clus

2.72

Disk

2.70

 Guilty

2.69

oqu

2.68

typ

2.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.