INDEX

Explanations

mentions of former positions or titles, or descriptions of individuals in news articles

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

phrases and terms related to political or historical events, especially those involving leadership and authority

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-att-kk

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

0:0.04

1:0.09

2:0.16

3:0.08

4:0.05

5:0.08

6:0.07

7:0.08

8:0.13

9:0.06

10:0.07

11:0.05

Negative Logits

 repetitive

-1.30

itaire

-1.23

 disag

-1.16

 curve

-1.16

 stale

-1.14

 demoral

-1.13

 unpredict

-1.11

definition

-1.10

FTWARE

-1.09

verage

-1.09

POSITIVE LOGITS

BuyableInstoreAndOnline

1.32

 Toledo

1.24

 extraord

1.20

Sag

1.17

 Strateg

1.15

 Alloy

1.10

 Adviser

1.10

 Medina

1.09

 Supervisor

1.08

 Founding

1.08

Activations Density 0.211%

mentions of former positions or titles, or descriptions of individuals in news articles

phrases and terms related to political or historical events, especially those involving leadership and authority

No Comments

No Known Activations

mentions of former positions or titles, or descriptions of individuals in news articles

phrases and terms related to political or historical events, especially those involving leadership and authority

No Comments

No Known Activations