INDEX

Explanations

mentions of geographical locations or landmarks

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.02

1:0.02

2:0.06

3:0.05

4:0.04

5:0.03

6:0.39

7:0.11

8:0.03

9:0.05

10:0.06

11:0.07

Negative Logits

 irregularities

-1.28

 autos

-1.17

 billboards

-1.16

 stitching

-1.14

 GOODMAN

-1.14

 ultrasound

-1.13

 violations

-1.13

 paran

-1.11

ATM

-1.11

 advertisements

-1.09

POSITIVE LOGITS

 Revenge

1.40

essen

1.38

]}

1.36

 Recover

1.35

 Ruin

1.33

 Romance

1.32

oris

1.24

 Despair

1.24

orgetown

1.22

riet

1.21

Activations Density 0.002%

mentions of geographical locations or landmarks

No Comments

No Known Activations