INDEX

Explanations

numerous mentions or references to works by different authors

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

Head Attr Weights

0:0.04

1:0.03

2:0.03

3:0.08

4:0.11

5:0.41

6:0.02

7:0.05

8:0.07

9:0.01

10:0.03

11:0.06

Negative Logits

DVD

-3.68

 Movie

-3.55

 subtitles

-3.51

 spoilers

-3.49

 Cinem

-3.48

 DVDs

-3.31

 documentaries

-3.28

 movie

-3.21

 Vengeance

-3.16

 Ghosts

-3.16

POSITIVE LOGITS

 unregulated

2.85

opes

2.82

ntil

2.82

ilateral

2.75

omedical

2.71

onomous

2.66

mediated

2.64

opian

2.63

 systematically

2.58

 bullied

2.56

Activations Density 0.101%

numerous mentions or references to works by different authors

No Known Activations