INDEX

Explanations

references to directional positions, particularly "left" and "right"

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_L9_Hcat_z_lr1.20e-03_l11.20e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.9.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

0:0.07

1:0.02

2:0.04

3:0.17

4:0.04

5:0.31

6:0.03

7:0.06

8:0.03

9:0.03

10:0.09

11:0.04

Negative Logits

 paralysis

-2.41

 strugg

-2.33

 moaning

-2.26

 slows

-2.24

pmwiki

-2.21

 nerves

-2.20

 drifting

-2.20

arrow

-2.15

 legs

-2.10

wings

-2.06

POSITIVE LOGITS

 GMOs

2.71

 Protocol

2.61

 Nikol

2.45

Eug

2.43

 Anthropology

2.41

 Preservation

2.39

 Recipe

2.33

 UNESCO

2.28

IPM

2.27

 Pharmac

2.25

Activations Density 0.015%

references to directional positions, particularly "left" and "right"

No Comments

No Known Activations