INDEX

Explanations

instructions for downloading files and versions, as well as SHA256 checksums for file verification

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

0:0.05

1:0.10

2:0.07

3:0.08

4:0.05

5:0.12

6:0.09

7:0.05

8:0.06

9:0.12

10:0.10

11:0.04

Negative Logits

-1.29

 powerless

-1.20

 destroyed

-1.18

 existence

-1.13

 therefore

-1.13

 Rousse

-1.12

 abolished

-1.11

 Galile

-1.11

 subordinates

-1.11

 hers

-1.11

POSITIVE LOGITS

PDATE

1.74

PDATED

1.60

FORMATION

1.46

Scroll

1.46

 Website

1.46

WARE

1.43

 Updates

1.42

Detailed

1.42

 Patreon

1.40

Click

1.39

Activations Density 0.314%

instructions for downloading files and versions, as well as SHA256 checksums for file verification

No Known Activations