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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.09

1:0.05

2:0.09

3:0.08

4:0.07

5:0.08

6:0.08

7:0.08

8:0.10

9:0.08

10:0.08

11:0.08

Negative Logits

accompan

-2.01

ciation

-1.70

inct

-1.63

 intrins

-1.62

 corrid

-1.61

 incomp

-1.60

 imped

-1.52

fix

-1.51

 conclud

-1.50

changes

-1.49

POSITIVE LOGITS

Ni

1.86

 Wolfgang

1.77

Rez

1.76

 Roof

1.68

Zur

1.63

 Totally

1.62

 Pioneer

1.59

 Zhao

1.58

 Shim

1.58

Pie

1.54

Activations Density 0.000%

No Known Activations

This feature has no known activations.