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

Head Attr Weights

0:0.08

1:0.08

2:0.09

3:0.09

4:0.08

5:0.08

6:0.06

7:0.07

8:0.08

9:0.07

10:0.08

11:0.09

Negative Logits

 Adult

-2.08

 Deadline

-1.69

 FILE

-1.68

 sketches

-1.62

 investigates

-1.58

CSI

-1.56

 Apprentice

-1.54

 Transgender

-1.53

 Reporter

-1.52

 ENTER

-1.52

POSITIVE LOGITS

soType

2.04

oslav

1.85

joy

1.79

prone

1.73

 horm

1.71

fu

1.70

sym

1.67

 blight

1.66

 Aren

1.62

addons

1.60

Activations Density 0.000%

No Known Activations

This feature has no known activations.