INDEX

Explanations

phrases indicating timelines or sequences of events

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

2:0.06

3:0.07

4:0.08

5:0.02

6:0.07

7:0.43

8:0.03

9:0.03

10:0.07

11:0.07

Negative Logits

bery

-1.67

berg

-1.65

othing

-1.61

paper

-1.58

し

-1.53

wine

-1.52

�

-1.45

aldo

-1.43

ervation

-1.42

bern

-1.41

POSITIVE LOGITS

 engagements

1.91

 completion

1.87

 unfolding

1.85

 events

1.83

 timelines

1.80

 rollout

1.78

 unfold

1.75

 timeline

1.72

 deployment

1.71

 initiation

1.71

Activations Density 0.010%

phrases indicating timelines or sequences of events

No Known Activations