INDEX

Explanations

terms related to communication or media

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

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L0_Hcat_z_lr1.20e-03_l11.80e+00_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.0.attn.hook_z

Hook Layer

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.02

1:0.15

2:0.03

3:0.16

4:0.16

5:0.16

6:0.03

7:0.05

8:0.06

9:0.03

10:0.08

11:0.01

Negative Logits

venge

-4.56

══

-4.30

execute

-4.24

gery

-4.12

ifling

-4.06

ACTED

-3.96

DonaldTrump

-3.95

integer

-3.92

href

-3.89

ⓘ

-3.89

POSITIVE LOGITS

ertodd

5.46

yip

5.41

terness

4.95

 interven

4.62

icter

4.54

emouth

4.41

nect

4.32

ombat

4.31

izoph

4.29

leground

4.29

Activations Density 0.002%

terms related to communication or media

No Comments

No Known Activations