INDEX

Explanations

phrases related to legal claims, court cases, and events

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Head Attr Weights

0:0.06

1:0.03

2:0.07

3:0.10

4:0.03

5:0.11

6:0.05

7:0.04

8:0.24

9:0.08

10:0.10

11:0.05

Negative Logits

arc

-1.23

ggles

-1.11

,...

-1.11

ships

-1.08

yrs

-1.05

)--

-1.04

kens

-1.04

FI

-1.03

ogue

-1.03

illon

-1.02

POSITIVE LOGITS

 裏�

1.31

cedented

1.25

 deserved

1.18

Notice

1.13

CLASSIFIED

1.13

evidence

1.11

 ingred

1.08

 TAMADRA

1.07

DERR

1.04

Bus

1.02

Activations Density 0.077%

phrases related to legal claims, court cases, and events

No Known Activations