INDEX

Explanations

words related to geographical locations and names

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

0:0.04

1:0.04

2:0.06

3:0.30

4:0.04

5:0.02

6:0.08

7:0.12

8:0.03

9:0.09

10:0.05

11:0.09

Negative Logits

emis

-1.34

ール

-1.32

anwhile

-1.21

 moratorium

-1.20

 hosp

-1.19

ukemia

-1.19

terday

-1.17

dule

-1.15

}}}

-1.14

 contrace

-1.14

POSITIVE LOGITS

 Duck

1.27

 Legends

1.18

ijk

1.16

 Borders

1.13

 Species

1.12

 Sheep

1.08

estone

1.08

 Lite

1.08

 Cotton

1.08

 Item

1.07

Activations Density 0.003%

words related to geographical locations and names

No Known Activations