INDEX

Explanations

numerical rankings or ordinal positions

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.07

1:0.08

2:0.07

3:0.08

4:0.07

5:0.09

6:0.07

7:0.07

8:0.07

9:0.08

10:0.09

11:0.08

Negative Logits

Uri

-1.70

 Euph

-1.59

 Galile

-1.55

detail

-1.49

 Juda

-1.43

razil

-1.40

 Sabha

-1.35

ディ

-1.34

 Tent

-1.33

SPONSORED

-1.32

POSITIVE LOGITS

croft

1.59

tut

1.49

izens

1.46

rous

1.45

龍�

1.44

 genius

1.42

uries

1.39

enberg

1.39

erb

1.38

 haun

1.37

Activations Density 0.000%

numerical rankings or ordinal positions

No Comments

No Known Activations