INDEX

Explanations

No Explanations Found

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.09

1:0.06

2:0.09

3:0.08

4:0.09

5:0.08

6:0.07

7:0.07

8:0.07

9:0.07

10:0.09

11:0.09

Negative Logits

 decriminal

-1.88

 sanitation

-1.74

 antibiotics

-1.66

 privat

-1.66

 rubbish

-1.62

 staples

-1.61

 detainees

-1.60

 shel

-1.54

 compromises

-1.54

netflix

-1.51

POSITIVE LOGITS

 Beir

2.16

PAN

1.85

 Manz

1.79

 Pere

1.79

McA

1.78

 Particip

1.75

 Nath

1.74

 Archangel

1.70

Gle

1.70

Ong

1.68

Activations Density 0.000%

No Known Activations

This feature has no known activations.