INDEX

Explanations

specific symbolic representations or quantifiers in a dataset

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

0:0.05

1:0.02

2:0.07

3:0.05

4:0.05

5:0.03

6:0.37

7:0.03

8:0.04

9:0.04

10:0.10

11:0.08

Negative Logits

 Fargo

-1.58

 Beir

-1.56

 Immunity

-1.34

 Swanson

-1.34

 dividends

-1.31

staking

-1.30

heit

-1.25

 Mankind

-1.24

 Myster

-1.23

 Mercy

-1.21

POSITIVE LOGITS

cas

1.82

ROM

1.55

asus

1.51

hang

1.51

odon

1.43

console

1.37

cheat

1.36

basketball

1.35

ESA

1.35

kit

1.32

Activations Density 0.002%

specific symbolic representations or quantifiers in a dataset

No Known Activations