INDEX

Explanations

references to violence and crime incidents

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

Not in Any Lists

No Comments

Head Attr Weights

0:0.14

1:0.03

2:0.10

3:0.08

4:0.07

5:0.09

6:0.18

7:0.03

8:0.04

9:0.06

10:0.07

11:0.07

Negative Logits

wolves

-1.72

esp

-1.60

cot

-1.48

quartered

-1.46

orage

-1.45

TEXTURE

-1.43

rays

-1.42

lyak

-1.42

bery

-1.41

ithing

-1.38

POSITIVE LOGITS

 REUTERS

1.55

 pause

1.54

 WATCHED

1.51

 WEEK

1.41

 NEWS

1.40

 Replay

1.39

IDA

1.39

 Advertisement

1.38

 PRESS

1.35

 hello

1.33

Activations Density 0.001%

references to violence and crime incidents

No Comments

No Known Activations