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

0:0.07

1:0.07

2:0.10

3:0.07

4:0.08

5:0.09

6:0.07

7:0.08

8:0.07

9:0.08

10:0.08

11:0.08

Negative Logits

 educ

-2.68

__

-2.48

 bureaucr

-2.42

 seiz

-2.41

 ����

-2.40

=-=-=-=-

-2.39

educ

-2.39

 ignor

-2.37

respons

-2.36

 dictators

-2.36

POSITIVE LOGITS

BST

2.95

 metre

2.83

 Boyle

2.63

RTX

2.58

AME

2.57

actionDate

2.51

 fortnight

2.48

 Melee

2.45

 Halifax

2.39

DH

2.37

Activations Density 0.000%

No Known Activations

This feature has no known activations.