INDEX

Explanations

details about news articles, such as images, captions, and specific dates

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

IFrame

Link

Head Attr Weights

0:0.05

1:0.10

2:0.04

3:0.04

4:0.03

5:0.06

6:0.07

7:0.13

8:0.06

9:0.13

10:0.16

11:0.07

Negative Logits

chance

-1.52

immune

-1.34

 attrition

-1.31

��

-1.29

volent

-1.25

theless

-1.25

 volunt

-1.24

 elig

-1.19

��

-1.18

cipled

-1.18

POSITIVE LOGITS

 Photos

1.77

 photos

1.72

umbnails

1.69

Pict

1.63

 pics

1.60

phot

1.56

Phot

1.56

photos

1.55

 photographers

1.53

 images

1.50

Activations Density 0.125%

details about news articles, such as images, captions, and specific dates

No Known Activations