INDEX

Explanations

instances of the phrase "follow up" or related terms that indicate continuation or connection to previous content

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

1:0.01

2:0.05

3:0.04

4:0.14

5:0.02

6:0.03

7:0.45

8:0.02

9:0.04

10:0.07

11:0.06

Negative Logits

erity

-1.58

Reviewer

-1.51

adem

-1.50

vantage

-1.43

agogue

-1.42

iltr

-1.41

ustomed

-1.39

twitch

-1.38

ortion

-1.37

 maneu

-1.33

POSITIVE LOGITS

meier

1.83

 previous

1.51

ensis

1.49

dated

1.45

dies

1.45

 unfinished

1.44

iago

1.41

 Notes

1.36

series

1.36

 defunct

1.35

Activations Density 0.002%

instances of the phrase "follow up" or related terms that indicate continuation or connection to previous content

No Known Activations