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

0:0.07

1:0.06

2:0.08

3:0.10

4:0.09

5:0.07

6:0.07

7:0.08

8:0.09

9:0.07

10:0.08

11:0.08

Negative Logits

 telev

-1.94

 sidel

-1.82

 playback

-1.79

 prosec

-1.79

 eleph

-1.72

owsky

-1.65

 Tibetan

-1.61

 hears

-1.56

 earthqu

-1.55

 proble

-1.54

POSITIVE LOGITS

omics

2.04

urgy

2.00

iology

1.80

 Masquerade

1.76

smanship

1.70

Games

1.67

isans

1.60

below

1.59

odan

1.56

isites

1.55

Activations Density 0.000%

No Known Activations

This feature has no known activations.