INDEX

Explanations

phrases indicating likelihood and probabilities of events or conditions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

Not in Any Lists

No Comments

Head Attr Weights

0:0.04

1:0.01

2:0.11

3:0.06

4:0.07

5:0.05

6:0.04

7:0.22

8:0.03

9:0.06

10:0.11

11:0.16

Negative Logits

calling

-1.38

worldly

-1.37

 senses

-1.30

flation

-1.29

 mund

-1.25

frog

-1.25

ollar

-1.23

beh

-1.23

leaf

-1.23

timer

-1.21

POSITIVE LOGITS

 casualties

1.26

 backups

1.24

headers

1.20

wre

1.20

 installing

1.19

 uncomp

1.18

 destroying

1.16

 obtaining

1.15

querque

1.13

semble

1.13

Activations Density 0.015%

phrases indicating likelihood and probabilities of events or conditions

No Comments

No Known Activations