INDEX

Explanations

No Explanations Found

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.07

1:0.06

2:0.06

3:0.13

4:0.05

5:0.06

6:0.06

7:0.11

8:0.06

9:0.07

10:0.12

11:0.09

Negative Logits

apor

-1.21

 infiltr

-1.21

 Emails

-1.17

 anyways

-1.15

 intimid

-1.06

aways

-1.05

 exacerb

-1.04

DragonMagazine

-1.04

 abound

-1.02

 bribes

-1.01

POSITIVE LOGITS

oso

1.22

ongyang

1.19

 certific

1.13

equal

1.13

aughtered

1.09

accompanied

1.07

ixt

1.06

idium

1.04

Recipe

1.03

anian

1.02

Activations Density 0.000%

No Known Activations

This feature has no known activations.