INDEX

Explanations

sentences containing numerical values or statistics

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

Head Attr Weights

0:0.05

1:0.11

2:0.04

3:0.05

4:0.05

5:0.26

6:0.03

7:0.04

8:0.05

9:0.23

10:0.03

11:0.03

Negative Logits

 Rather

-2.72

 Worse

-2.66

 Thus

-2.51

 unsuspecting

-2.49

Furthermore

-2.43

Moreover

-2.37

Thus

-2.36

 lowly

-2.35

 Instead

-2.34

 Alas

-2.34

POSITIVE LOGITS

 VIDE

3.27

 actionGroup

2.86

 vibe

2.45

GMT

2.38

 gotta

2.23

>[

2.20

GMT

2.18

 kinda

2.17

 maybe

2.16

)",

2.14

Activations Density 0.017%

sentences containing numerical values or statistics

No Known Activations