INDEX

Explanations

references to locations, specifically cities and towns

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

0:0.06

1:0.02

2:0.19

3:0.05

4:0.09

5:0.07

6:0.04

7:0.04

8:0.13

9:0.06

10:0.14

11:0.05

Negative Logits

erity

-1.71

 sidx

-1.68

oos

-1.66

uctor

-1.65

afort

-1.63

ividual

-1.60

pload

-1.60

 spouses

-1.56

letes

-1.56

 necks

-1.52

POSITIVE LOGITS

 birthplace

1.88

 Dragonbound

1.79

 inhabited

1.78

yards

1.67

 neighbouring

1.66

esi

1.66

 Seraph

1.64

 Aram

1.64

\<

1.63

[|

1.62

Activations Density 0.011%

references to locations, specifically cities and towns

No Known Activations