INDEX

Explanations

phrases referring to individuals and their opinions or actions

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

Head Attr Weights

0:0.02

1:0.03

2:0.07

3:0.31

4:0.06

5:0.04

6:0.08

7:0.05

8:0.04

9:0.03

10:0.12

11:0.09

Negative Logits

 Debor

-1.73

 Frey

-1.64

 Beir

-1.57

 Buenos

-1.51

 Paso

-1.50

 Erica

-1.50

 Kear

-1.47

 Chou

-1.43

 Sach

-1.40

 Sieg

-1.39

POSITIVE LOGITS

raviolet

1.80

istries

1.78

azine

1.69

 Topic

1.67

yton

1.65

assian

1.61

Roaming

1.61

amins

1.61

obyl

1.60

atan

1.56

Activations Density 0.092%

phrases referring to individuals and their opinions or actions

No Known Activations