INDEX

Explanations

proper nouns and names

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_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.3.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.02

1:0.03

2:0.08

3:0.23

4:0.01

5:0.02

6:0.15

7:0.17

8:0.03

9:0.08

10:0.09

11:0.04

Negative Logits

dylib

-1.12

formation

-1.08

otomy

-1.06

Http

-1.05

DeL

-0.96

Peb

-0.95

 Bahamas

-0.95

 Float

-0.95

 Betty

-0.94

TEXTURE

-0.93

POSITIVE LOGITS

kinson

1.25

wer

1.16

meier

1.08

igan

1.08

将

1.05

戦

1.05

essen

1.05

alle

1.04

maxwell

1.04

verty

1.03

Activations Density 0.005%

proper nouns and names

No Known Activations