INDEX

Explanations

phrases related to expressing opinions or explanations

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

IFrame

Link

Not in Any Lists

No Comments

Head Attr Weights

0:0.11

1:0.05

2:0.10

3:0.11

4:0.06

5:0.09

6:0.06

7:0.04

8:0.13

9:0.11

10:0.08

11:0.03

Negative Logits

 shitty

-1.36

 crappy

-1.15

 messed

-1.13

 asshole

-1.11

 blah

-1.09

 endeavors

-1.09

 kinda

-1.08

 piss

-1.08

 bullshit

-1.08

 dude

-1.07

POSITIVE LOGITS

terday

1.33

CTV

1.32

enegger

1.23

 srfAttach

1.21

WAYS

1.18

rov

1.13

iod

1.11

�

1.10

 REUTERS

1.10

gans

1.09

Activations Density 0.037%

phrases related to expressing opinions or explanations

No Comments

No Known Activations