INDEX

Explanations

references to individuals involved in plots or wrongdoings

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

1:0.03

2:0.05

3:0.13

4:0.02

5:0.04

6:0.09

7:0.20

8:0.04

9:0.05

10:0.08

11:0.18

Negative Logits

ustomed

-1.35

enhagen

-1.19

interrupted

-1.15

 preview

-1.14

isSpecialOrderable

-1.11

izons

-1.08

 mornings

-1.08

owship

-1.06

 snapshots

-1.05

ciating

-1.05

POSITIVE LOGITS

 Kinnikuman

1.17

jee

1.13

�

1.12

awar

1.11

 rigging

1.09

MAD

1.08

 Mohammad

1.04

{}

1.04

 villain

1.03

CPR

1.02

Activations Density 0.004%

references to individuals involved in plots or wrongdoings

No Comments

No Known Activations