INDEX

Explanations

references to political contexts and terms

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

0:0.02

1:0.02

2:0.05

3:0.06

4:0.05

5:0.04

6:0.45

7:0.04

8:0.05

9:0.06

10:0.07

11:0.05

Negative Logits

 spoilers

-1.26

------------------------------------------------

-1.21

 wonder

-1.18

 specificity

-1.17

 comprehension

-1.16

ension

-1.12

 consideration

-1.12

Winter

-1.11

olesterol

-1.11

ended

-1.08

POSITIVE LOGITS

geist

1.62

neau

1.45

��

1.39

aceae

1.35

zyk

1.35

�

1.33

clinton

1.31

ça

1.28

wagen

1.26

veland

1.25

Activations Density 0.003%

references to political contexts and terms

No Comments

No Known Activations