INDEX

Explanations

references to speakers and discussions in events or panels

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_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

0:0.03

1:0.02

2:0.05

3:0.05

4:0.05

5:0.04

6:0.04

7:0.06

8:0.03

9:0.06

10:0.39

11:0.13

Negative Logits

 regulate

-1.85

 enforced

-1.83

 inherited

-1.77

 regulates

-1.76

 regulatory

-1.69

FSA

-1.68

 regulated

-1.64

 regulating

-1.64

 policing

-1.59

 enforce

-1.58

POSITIVE LOGITS

Panel

1.69

 sidx

1.57

stars

1.55

 Tent

1.52

Stars

1.52

Ent

1.51

Bob

1.48

 Emin

1.48

Flo

1.47

 Anderson

1.46

Activations Density 0.018%

references to speakers and discussions in events or panels

No Comments

No Known Activations