INDEX

Explanations

specific words or phrases related to instructions or steps to follow

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

1:0.02

2:0.14

3:0.04

4:0.31

5:0.04

6:0.02

7:0.02

8:0.17

9:0.06

10:0.05

11:0.02

Negative Logits

 depend

-1.53

ranged

-1.51

efer

-1.39

 dependence

-1.35

depending

-1.31

zie

-1.28

song

-1.26

ank

-1.24

grand

-1.21

 bragging

-1.20

POSITIVE LOGITS

 balances

1.36

 Dictionary

1.32

 positives

1.29

anwhile

1.27

 captcha

1.27

POL

1.25

 sidelines

1.25

 Janeiro

1.24

abus

1.21

irmed

1.21

Activations Density 0.023%

specific words or phrases related to instructions or steps to follow

No Comments

No Known Activations