INDEX

Explanations

phrases indicating presence or position in relation to objects or concepts

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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.02

1:0.01

2:0.13

3:0.06

4:0.13

5:0.03

6:0.04

7:0.37

8:0.02

9:0.03

10:0.05

11:0.06

Negative Logits

soever

-1.75

oral

-1.56

icide

-1.55

 curfew

-1.55

TION

-1.48

 forcefully

-1.47

onal

-1.47

rompt

-1.47

uncture

-1.44

 deepen

-1.44

POSITIVE LOGITS

 pedest

1.68

 Stras

1.59

 shelves

1.58

 throne

1.58

 Drac

1.54

 charts

1.49

 Saur

1.49

 heaven

1.48

 Satoshi

1.44

uggage

1.43

Activations Density 0.001%

phrases indicating presence or position in relation to objects or concepts

No Comments

No Known Activations