INDEX

Explanations

names and initials associated with individuals, particularly in a specific context or narrative

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

0:0.03

1:0.01

2:0.04

3:0.05

4:0.05

5:0.03

6:0.37

7:0.16

8:0.04

9:0.05

10:0.07

11:0.04

Negative Logits

antically

-1.47

 Kare

-1.33

ּ

-1.32

ovi

-1.19

UA

-1.19

atically

-1.15

orage

-1.13

 Galile

-1.11

spons

-1.10

 Scientology

-1.10

POSITIVE LOGITS

brook

1.45

Winged

1.36

ESPN

1.35

snap

1.34

emouth

1.30

CNN

1.30

steen

1.29

SHA

1.28

Rain

1.25

idge

1.25

Activations Density 0.002%

names and initials associated with individuals, particularly in a specific context or narrative

No Comments

No Known Activations