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

0:0.08

1:0.08

2:0.07

3:0.09

4:0.09

5:0.07

6:0.08

7:0.07

8:0.07

9:0.08

10:0.08

11:0.08

Negative Logits

 antibodies

-2.49

 Furious

-2.47

Klu

-2.43

 Shots

-2.34

 chemicals

-2.33

 Feeling

-2.32

aghan

-2.31

 Pipeline

-2.27

 Gamble

-2.26

 juices

-2.25

POSITIVE LOGITS

jri

2.67

obook

2.61

rament

2.56

Nib

2.52

ardless

2.52

ikk

2.40

soDeliveryDate

2.40

DragonMagazine

2.38

ansk

2.37

 Cain

2.37

Activations Density 0.000%

No Known Activations

This feature has no known activations.