INDEX

Explanations

headlines and mentions of reports from Colliers International

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

0:0.01

1:0.04

2:0.09

3:0.06

4:0.02

5:0.05

6:0.09

7:0.20

8:0.09

9:0.08

10:0.06

11:0.14

Negative Logits

ć

-1.43

 imperfect

-1.21

 labou

-1.20

 improperly

-1.17

 cataly

-1.16

 compress

-1.12

 implanted

-1.12

 laborers

-1.10

 entangled

-1.09

 rigged

-1.09

POSITIVE LOGITS

tainment

1.47

News

1.47

TRUMP

1.44

 Editorial

1.34

Quotes

1.33

Sad

1.27

OPLE

1.27

の�

1.23

USA

1.22

Talking

1.19

Activations Density 0.002%

headlines and mentions of reports from Colliers International

No Known Activations