INDEX

Explanations

names of people and organizations, particularly their specific titles or roles

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

identifiers for various car manufacturers and models

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-att-kk

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.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

1:0.04

2:0.26

3:0.08

4:0.14

5:0.06

6:0.03

7:0.03

8:0.05

9:0.10

10:0.07

11:0.03

Negative Logits

 Barbarian

-1.44

 Races

-1.36

icularly

-1.18

seekers

-1.15

 Appalachian

-1.11

 Caucus

-1.05

 Appalach

-1.03

nder

-1.03

ça

-1.03

 Archie

-1.01

POSITIVE LOGITS

wagen

1.43

Sov

1.33

pload

1.33

obiles

1.23

hiba

1.21

ynt

1.20

vic

1.19

otive

1.17

vu

1.15

ueller

1.15

Activations Density 0.001%

names of people and organizations, particularly their specific titles or roles

identifiers for various car manufacturers and models

No Comments

No Known Activations