INDEX

Explanations

No Explanations Found

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

Not in Any Lists

No Comments

Head Attr Weights

0:0.09

1:0.12

2:0.07

3:0.07

4:0.05

5:0.07

6:0.09

7:0.08

8:0.07

9:0.07

10:0.12

11:0.06

Negative Logits

 blah

-1.66

[/

-1.41

Yep

-1.35

Oh

-1.34

Yep

-1.33

 anyways

-1.31

malink

-1.30

 EVERY

-1.30

 Yeah

-1.29

 asshole

-1.28

POSITIVE LOGITS

 Kazakh

1.75

 manoeuv

1.63

 Nato

1.46

awei

1.40

 diplomatic

1.40

 parliamentary

1.38

 Kazakhstan

1.37

 diplomats

1.36

ilateral

1.36

assian

1.35

Activations Density 0.000%

No Known Activations

This feature has no known activations.