INDEX

Explanations

negative terms related to unjust investigations or trials

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_L2_Hcat_z_lr1.20e-03_l11.00e+00_ds24576_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v4.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

float32

Hook Name

blocks.2.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

0:0.04

1:0.03

2:0.24

3:0.06

4:0.17

5:0.06

6:0.02

7:0.03

8:0.06

9:0.17

10:0.05

11:0.02

Negative Logits

PsyNetMessage

-1.33

andowski

-1.30

 differential

-1.29

 Calories

-1.29

 GOODMAN

-1.28

 reinforcement

-1.23

 Adinida

-1.21

 stabilization

-1.21

 upstream

-1.18

Sensor

-1.17

POSITIVE LOGITS

lda

1.58

doctor

1.42

 Madagascar

1.33

 Monstrous

1.31

ゴン

1.30

 fame

1.21

Anne

1.21

Sue

1.20

 Peninsula

1.20

andering

1.19

Activations Density 0.005%

negative terms related to unjust investigations or trials

No Comments

No Known Activations