INDEX

Explanations

political campaign-related phrases and terms

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

phrases related to campaign strategies and endorsements

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

0:0.02

1:0.03

2:0.06

3:0.03

4:0.02

5:0.07

6:0.13

7:0.36

8:0.07

9:0.02

10:0.06

11:0.07

Negative Logits

veland

-1.21

otropic

-1.17

 disapp

-1.14

 dehyd

-1.13

iths

-1.11

ensual

-1.09

ertodd

-1.09

 coerc

-1.04

igent

-1.04

olithic

-1.03

POSITIVE LOGITS

govtrack

1.37

 billed

1.26

 headlined

1.26

andise

1.25

quickShipAvailable

1.24

 Hollande

1.22

 chaired

1.22

RAFT

1.18

 Hannity

1.17

 dubbed

1.12

Activations Density 0.059%

political campaign-related phrases and terms

phrases related to campaign strategies and endorsements

No Comments

No Known Activations