INDEX

Explanations

discussions about the advantages and disadvantages of various topics or decisions

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

0:0.03

1:0.01

2:0.05

3:0.06

4:0.08

5:0.04

6:0.05

7:0.38

8:0.04

9:0.05

10:0.09

11:0.07

Negative Logits

roots

-1.81

LES

-1.61

 forearm

-1.57

hem

-1.54

REF

-1.47

MRI

-1.45

NH

-1.42

 eyebrow

-1.38

umblr

-1.35

root

-1.35

POSITIVE LOGITS

 achie

1.66

 outweigh

1.59

 econom

1.56

 Azerb

1.56

 aspiring

1.54

 Surviv

1.53

 prospects

1.53

achev

1.49

 ranking

1.47

 compan

1.46

Activations Density 0.002%

discussions about the advantages and disadvantages of various topics or decisions

No Known Activations