INDEX

Explanations

the concept of distance or distance-related ideas

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_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.3.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

2:0.06

3:0.11

4:0.01

5:0.04

6:0.12

7:0.17

8:0.06

9:0.09

10:0.07

11:0.18

Negative Logits

awa

-1.19

ullivan

-1.16

 accomp

-1.09

yss

-1.08

 enthusiastically

-1.06

wagen

-1.05

 Marketable

-1.05

OUP

-1.04

 rhet

-1.03

 Orche

-1.03

POSITIVE LOGITS

 Territories

1.41

uras

1.21

 destinations

1.20

 vantage

1.15

 destination

1.15

rome

1.14

mid

1.13

istrate

1.08

addr

1.06

hend

1.06

Activations Density 0.004%

the concept of distance or distance-related ideas

No Known Activations