INDEX

Explanations

phrases and words associated with challenges and difficulties

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.3.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

1:0.03

2:0.06

3:0.20

4:0.02

5:0.03

6:0.06

7:0.20

8:0.06

9:0.08

10:0.07

11:0.12

Negative Logits

soDeliveryDate

-1.53

ther

-1.38

=-=-=-=-

-1.21

ledged

-1.20

ETH

-1.15

�

-1.15

Jer

-1.14

Bet

-1.12

angel

-1.12

フォ

-1.11

POSITIVE LOGITS

 digestion

1.29

 obtaining

1.14

 maneuver

1.14

 acquiring

1.11

 acquisition

1.09

 penetrate

1.07

 penetrating

1.06

 handling

1.05

 attain

1.04

 math

1.03

Activations Density 0.006%

phrases and words associated with challenges and difficulties

No Known Activations