INDEX

Explanations

phrases indicating purpose or intention

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

places and structures

oai_token-act-pair · gpt-3.5-turbo

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

0:0.03

1:0.02

2:0.15

3:0.03

4:0.31

5:0.09

6:0.03

7:0.02

8:0.08

9:0.13

10:0.04

11:0.02

Negative Logits

otaur

-1.64

assian

-1.57

illusion

-1.54

netflix

-1.46

ioxide

-1.46

problem

-1.45

nexus

-1.44

erm

-1.44

aho

-1.43

abe

-1.42

POSITIVE LOGITS

 destro

1.64

 derog

1.50

 punishing

1.46

 Sacrifice

1.36

 arrang

1.27

 storing

1.25

 charitable

1.25

FUL

1.25

 conclud

1.25

 referen

1.22

Activations Density 0.008%

phrases indicating purpose or intention

places and structures

No Known Activations