INDEX

Explanations

suffixes that indicate actions or states

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.03

1:0.03

2:0.05

3:0.06

4:0.05

5:0.04

6:0.36

7:0.04

8:0.05

9:0.05

10:0.08

11:0.10

Negative Logits

 brakes

-1.42

govtrack

-1.41

 inappropriate

-1.36

 Awareness

-1.35

 inaction

-1.34

agna

-1.33

ा

-1.24

 neglect

-1.24

ptroller

-1.22

IPM

-1.22

POSITIVE LOGITS

Secure

1.39

ometown

1.32

glers

1.25

venth

1.23

riot

1.21

Ct

1.19

 occasions

1.18

 Legions

1.17

deck

1.17

upon

1.17

Activations Density 0.001%

suffixes that indicate actions or states

No Comments

No Known Activations