INDEX

Explanations

phrases related to emotional sensitivity or fragility

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L5_Hcat_z_lr1.20e-03_l11.00e+00_ds49152_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

float32

Hook Name

blocks.5.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.01

1:0.01

2:0.12

3:0.07

4:0.08

5:0.03

6:0.21

7:0.21

8:0.03

9:0.04

10:0.06

11:0.07

Negative Logits

ा

-1.38

 includ

-1.36

ゴン

-1.35

etheus

-1.35

 challeng

-1.34

wic

-1.33

untarily

-1.31

 accordingly

-1.30

 aptly

-1.28

anwhile

-1.28

POSITIVE LOGITS

 anymore

1.69

nor

1.62

 torches

1.28

nor

1.26

thin

1.26

 faint

1.25

 morbid

1.25

tiny

1.24

scl

1.22

abal

1.22

Activations Density 0.000%

phrases related to emotional sensitivity or fragility

No Known Activations