INDEX

Explanations

names, particularly those related to historical figures

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

New Auto-Interp

Top Features by Cosine Similarity

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.06

1:0.03

2:0.40

3:0.08

4:0.05

5:0.05

6:0.03

7:0.03

8:0.05

9:0.09

10:0.05

11:0.03

Negative Logits

���

-1.56

��

-1.32

��

-1.28

kefeller

-1.28

 Citiz

-1.24

 comr

-1.20

 pestic

-1.20

pps

-1.18

 Pascal

-1.15

\\\\\\\\

-1.14

POSITIVE LOGITS

iland

1.49

addin

1.29

amed

1.28

ilib

1.26

achus

1.25

inho

1.24

yi

1.19

cery

1.19

essim

1.18

ourning

1.18

Activations Density 0.006%

names, particularly those related to historical figures

No Comments

No Known Activations