INDEX

Explanations

No Explanations Found

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

0:0.08

1:0.05

2:0.08

3:0.09

4:0.09

5:0.08

6:0.07

7:0.08

8:0.08

9:0.10

10:0.07

11:0.07

Negative Logits

 Ples

-2.21

 Lumpur

-1.83

 Slug

-1.81

\/\/

-1.80

htaking

-1.74

okia

-1.74

ichita

-1.74

ubes

-1.72

arnaev

-1.71

uminati

-1.70

POSITIVE LOGITS

 returns

1.55

uitive

1.45

 consolation

1.45

 reass

1.42

―

1.40

 subt

1.38

 comp

1.38

Guy

1.33

 Gavin

1.31

Gu

1.30

Activations Density 0.000%

No Known Activations

This feature has no known activations.