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

1:0.06

2:0.07

3:0.08

4:0.07

5:0.09

6:0.07

7:0.08

8:0.09

9:0.10

10:0.08

11:0.08

Negative Logits

ネ

-1.75

 Alone

-1.73

 Commands

-1.72

 Freed

-1.71

 Aging

-1.67

anqu

-1.67

 Command

-1.67

 Shogun

-1.60

 Andromeda

-1.59

 Polic

-1.59

POSITIVE LOGITS

netflix

2.13

Reviewer

1.83

bum

1.82

yz

1.81

redients

1.78

aston

1.70

ylan

1.66

"}],"

1.59

yg

1.57

ribune

1.55

Activations Density 0.000%

No Known Activations

This feature has no known activations.