INDEX

Explanations

words and characters related to specific names or brands

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

0:0.06

1:0.03

2:0.05

3:0.09

4:0.02

5:0.04

6:0.14

7:0.02

8:0.03

9:0.42

10:0.03

11:0.03

Negative Logits

 Schultz

-4.52

 Shannon

-3.73

Kar

-3.63

 Hawk

-3.60

 Swan

-3.54

 Shawn

-3.54

 Shan

-3.54

JO

-3.29

 Russ

-3.29

Russ

-3.28

POSITIVE LOGITS

ode

5.89

ODE

5.66

odes

5.27

ome

5.15

omes

5.06

omer

4.46

OME

4.01

 decode

3.81

 Pompe

3.81

oded

3.78

Activations Density 0.001%

words and characters related to specific names or brands

No Comments

No Known Activations