INDEX

Explanations

frequent nouns and phrases that indicate rank or position

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

0:0.15

1:0.29

2:0.03

3:0.02

4:0.05

5:0.15

6:0.04

7:0.02

8:0.08

9:0.05

10:0.04

11:0.03

Negative Logits

ARK

-1.81

 Scrib

-1.75

ICE

-1.72

akery

-1.72

SC

-1.62

ZA

-1.62

scope

-1.61

CA

-1.61

esc

-1.61

CK

-1.58

POSITIVE LOGITS

 happiest

2.56

 richest

2.26

 brightest

2.18

 busiest

2.13

 strongest

2.05

 greatest

2.03

 wealthiest

2.02

 widest

2.01

liest

2.00

 deadliest

1.99

Activations Density 0.009%

frequent nouns and phrases that indicate rank or position

No Known Activations