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

1:0.08

2:0.09

3:0.08

4:0.06

5:0.10

6:0.08

7:0.07

8:0.08

9:0.08

10:0.07

11:0.08

Negative Logits

lycer

-1.75

hetics

-1.55

 commod

-1.51

agi

-1.46

 decom

-1.44

amel

-1.43

 accompl

-1.42

 Straw

-1.40

ortment

-1.40

tten

-1.38

POSITIVE LOGITS

Chart

1.95

VIEW

1.68

DAY

1.67

Offline

1.62

WATCH

1.61

biz

1.56

�

1.55

 BOOK

1.54

\/

1.52

INFO

1.48

Activations Density 0.000%

No Known Activations

This feature has no known activations.