INDEX

Explanations

actions or desires related to specific intentions or activities

oai_token-act-pair · gpt-3.5-turbo

Triggered by @bot

phrases indicating desires or intentions

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-att-kk

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

Head Attr Weights

0:0.07

1:0.03

2:0.11

3:0.10

4:0.12

5:0.07

6:0.04

7:0.03

8:0.20

9:0.09

10:0.05

11:0.02

Negative Logits

 Globe

-1.27

Laf

-1.23

 Vest

-1.16

 Glob

-1.14

 Fowler

-1.14

 Bloomberg

-1.09

 Annotations

-1.08

requires

-1.08

 Lions

-1.07

Sag

-1.06

POSITIVE LOGITS

thood

1.66

ドラ

1.59

 FANTASY

1.41

 hairc

1.32

icy

1.30

hetics

1.28

ロ

1.26

ュ

1.19

enance

1.17

IENT

1.16

Activations Density 0.007%

actions or desires related to specific intentions or activities

phrases indicating desires or intentions

No Known Activations