INDEX

Explanations

phrases indicating negative or cautionary scenarios, often linked to failures or challenges

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_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.3.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

1:0.02

2:0.11

3:0.07

4:0.02

5:0.02

6:0.15

7:0.11

8:0.07

9:0.05

10:0.05

11:0.24

Negative Logits

 wisely

-1.47

 Koran

-1.29

 boldly

-1.16

soever

-1.12

Yaz

-1.10

 diligently

-1.08

 loudly

-1.06

 Doodle

-1.06

Pru

-1.06

 Brune

-1.05

POSITIVE LOGITS

ensive

1.36

uclear

1.33

rontal

1.33

obyl

1.28

ricular

1.26

dozen

1.23

death

1.23

oval

1.22

termin

1.22

heric

1.22

Activations Density 0.024%

phrases indicating negative or cautionary scenarios, often linked to failures or challenges

No Comments

No Known Activations