INDEX

Explanations

comparisons between entities

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

1:0.06

2:0.05

3:0.08

4:0.03

5:0.10

6:0.07

7:0.08

8:0.20

9:0.08

10:0.11

11:0.05

Negative Logits

ledged

-1.61

 Cosponsors

-1.45

acebook

-1.33

aughtered

-1.30

arma

-1.28

rypted

-1.22

imeo

-1.22

icipated

-1.21

essage

-1.19

COMPLE

-1.17

POSITIVE LOGITS

 traditionally

1.31

 traditional

1.17

 typical

1.16

 Frieza

1.15

 conventional

1.15

 beginners

1.14

 purely

1.13

 inexperienced

1.12

 Aman

1.12

 dict

1.12

Activations Density 0.027%

comparisons between entities

No Known Activations