INDEX

Explanations

phrases that indicate exclusivity or uniqueness

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L9_Hcat_z_lr1.20e-03_l11.20e+00_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.9.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.07

1:0.09

2:0.08

3:0.08

4:0.09

5:0.08

6:0.07

7:0.08

8:0.08

9:0.08

10:0.07

11:0.07

Negative Logits

ño

-2.84

anqu

-2.68

CHA

-2.63

enne

-2.54

ictions

-2.53

acci

-2.47

neapolis

-2.45

achu

-2.44

 saint

-2.43

ampire

-2.40

POSITIVE LOGITS

 Branch

2.89

 McDonnell

2.87

 Fallon

2.49

 Dawson

2.49

 Thread

2.46

 Tide

2.45

 Applicant

2.44

 Redmond

2.43

 Proceedings

2.40

 Dart

2.39

Activations Density 0.000%

phrases that indicate exclusivity or uniqueness

No Comments

No Known Activations