INDEX

Explanations

phrases indicating connections or relationships, especially involving possession or descriptions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.02

1:0.02

2:0.07

3:0.05

4:0.07

5:0.02

6:0.25

7:0.28

8:0.03

9:0.03

10:0.07

11:0.05

Negative Logits

izons

-1.36

gdala

-1.30

erion

-1.29

orthy

-1.24

fml

-1.23

 Opportun

-1.22

Kad

-1.19

 dearly

-1.18

 nowhere

-1.18

fman

-1.17

POSITIVE LOGITS

anchester

1.59

unda

1.44

iform

1.43

 flowers

1.37

ridges

1.34

emetery

1.29

aign

1.29

 Flowers

1.24

avior

1.24

 demonstrations

1.24

Activations Density 0.001%

phrases indicating connections or relationships, especially involving possession or descriptions

No Comments

No Known Activations