INDEX

Explanations

phrases associated with authoritative statements and expert opinions

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

0:0.01

1:0.01

2:0.30

3:0.13

4:0.07

5:0.02

6:0.03

7:0.04

8:0.07

9:0.04

10:0.14

11:0.08

Negative Logits

swick

-1.35

 retake

-1.31

 seemed

-1.31

ortmund

-1.24

 totaled

-1.24

iversary

-1.23

iage

-1.22

 takeoff

-1.21

veyard

-1.19

 consisted

-1.18

POSITIVE LOGITS

anu

1.39

 acronym

1.26

Jah

1.20

 understatement

1.19

 jargon

1.17

dub

1.16

 persu

1.15

cience

1.13

 liar

1.11

 charism

1.10

Activations Density 0.075%

phrases associated with authoritative statements and expert opinions

No Comments

No Known Activations