INDEX

Explanations

phrases indicating subjective experiences or personal evaluations

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

0:0.02

1:0.02

2:0.06

3:0.07

4:0.19

5:0.03

6:0.07

7:0.30

8:0.04

9:0.03

10:0.06

11:0.05

Negative Logits

TING

-1.50

DonaldTrump

-1.44

istan

-1.42

soon

-1.39

Oracle

-1.37

lished

-1.36

answered

-1.30

uala

-1.30

 Nadu

-1.29

 Mehran

-1.25

POSITIVE LOGITS

 qualities

2.17

 traits

2.16

 characteristics

1.86

 strengths

1.84

 firepower

1.75

 weaknesses

1.68

 flexibility

1.66

natureconservancy

1.65

 Thumbnails

1.60

 behaviors

1.60

Activations Density 0.005%

phrases indicating subjective experiences or personal evaluations

No Known Activations