INDEX

Explanations

questions or phrases indicating inquiry and reflection

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L10_Hcat_z_lr1.20e-03_l11.30e+00_ds24576_bs4096_dc1.00e-05_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.10.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

1:0.16

2:0.08

3:0.03

4:0.05

5:0.02

6:0.06

7:0.12

8:0.03

9:0.04

10:0.12

11:0.08

Negative Logits

robe

-2.48

 curtains

-2.48

omnia

-2.46

 Genie

-2.46

 Moonlight

-2.44

azz

-2.36

uberty

-2.31

Dup

-2.29

 Elon

-2.25

Catal

-2.22

POSITIVE LOGITS

 Stick

5.82

 stick

5.68

 sticks

5.09

stick

4.80

sticks

4.25

 sticky

3.50

 sticking

3.31

 stuck

3.25

 glue

3.09

atan

2.88

Activations Density 0.006%

questions or phrases indicating inquiry and reflection

No Known Activations