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

IFrame

Link

Head Attr Weights

0:0.07

1:0.04

2:0.09

3:0.09

4:0.08

5:0.10

6:0.08

7:0.07

8:0.09

9:0.08

10:0.09

11:0.09

Negative Logits

 Cerberus

-1.85

 antitrust

-1.69

 Bethlehem

-1.47

 charms

-1.47

 Bethesda

-1.47

 perks

-1.46

 Comcast

-1.44

 accol

-1.42

 exemptions

-1.40

 unveiling

-1.39

POSITIVE LOGITS

ンジ

1.82

oad

1.74

1.60

田

1.57

ulnerable

1.54

フ

1.53

docker

1.53

ム

1.51

adic

1.51

aze

1.50

Activations Density 0.000%

No Known Activations

This feature has no known activations.