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

0:0.08

1:0.09

2:0.08

3:0.08

4:0.09

5:0.08

6:0.08

7:0.08

8:0.08

9:0.07

10:0.07

11:0.07

Negative Logits

DOD

-1.69

 Strat

-1.63

 retaliation

-1.56

 Return

-1.53

 meanwhile

-1.50

 Behavioral

-1.45

RECT

-1.42

 behavioral

-1.40

 unintended

-1.39

 premature

-1.39

POSITIVE LOGITS

bledon

2.06

antic

1.95

acas

1.91

apa

1.90

arse

1.83

ounce

1.78

antry

1.77

gran

1.74

abase

1.74

achus

1.69

Activations Density 0.000%

No Known Activations

This feature has no known activations.