INDEX

Explanations

phrases that indicate relationships or connections between entities

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

0:0.12

1:0.03

2:0.17

3:0.12

4:0.06

5:0.05

6:0.03

7:0.02

8:0.11

9:0.05

10:0.07

11:0.12

Negative Logits

%:

-1.65

!:

-1.61

mosp

-1.61

.:

-1.57

again

-1.54

Secondly

-1.51

Fed

-1.49

():

-1.49

ailable

-1.47

usual

-1.46

POSITIVE LOGITS

etc

1.91

 encyclopedia

1.65

...)

1.57

 timet

1.45

dq

1.43

…)

1.42

 submarine

1.38

 locom

1.36

 Nanto

1.35

 Trident

1.33

Activations Density 0.060%

phrases that indicate relationships or connections between entities

No Known Activations

phrases that indicate relationships or connections between entities

No Known Activations