INDEX

Explanations

technical or security-related features and capabilities of software

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_L10_Hcat_z_lr1.20e-03_l11.30e+00_ds24576_bs4096_dc1.00e-05_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.10.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

0:0.17

1:0.14

2:0.09

3:0.06

4:0.05

5:0.02

6:0.06

7:0.12

8:0.03

9:0.04

10:0.12

11:0.04

Negative Logits

 gays

-2.88

BCC

-2.87

gay

-2.87

tel

-2.86

Gay

-2.85

 Wallace

-2.83

 Havana

-2.78

 McCull

-2.77

 Lesbian

-2.70

grass

-2.69

POSITIVE LOGITS

 Epic

7.66

 Unreal

5.85

Ep

4.70

 Gears

3.77

UE

3.37

 Quake

3.31

Ep

3.30

 epic

3.28

EPS

3.22

 Ogre

3.20

Activations Density 0.002%

technical or security-related features and capabilities of software

No Comments

No Known Activations