INDEX

Explanations

negative constructs or phrases related to scrutiny and lack of accountability in various contexts

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

New Auto-Interp

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

0:0.06

1:0.01

2:0.19

3:0.20

4:0.12

5:0.03

6:0.06

7:0.03

8:0.10

9:0.03

10:0.05

11:0.07

Negative Logits

 anonym

-1.46

 icing

-1.43

ynchron

-1.39

 backlog

-1.38

tumblr

-1.38

 mash

-1.31

 imagining

-1.29

 slur

-1.27

 vaguely

-1.27

 ranging

-1.25

POSITIVE LOGITS

nor

3.54

nor

2.35

 anymore

2.20

inventoryQuantity

1.97

irlf

1.81

Nor

1.75

Nor

1.74

omsky

1.65

ught

1.64

ridges

1.56

Activations Density 0.019%

negative constructs or phrases related to scrutiny and lack of accountability in various contexts

No Comments

No Known Activations