INDEX

Explanations

instances of the words "top" and "bottom" related to position or placement

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

New Auto-Interp

Top Features by Cosine Similarity

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

1:0.03

2:0.11

3:0.09

4:0.02

5:0.05

6:0.13

7:0.07

8:0.11

9:0.05

10:0.14

11:0.10

Negative Logits

thood

-1.53

theless

-1.33

amaz

-1.31

 toast

-1.22

 forgiven

-1.22

ビ

-1.21

igmatic

-1.19

accompan

-1.16

netflix

-1.15

basketball

-1.14

POSITIVE LOGITS

 Consumer

1.43

rium

1.39

 snail

1.25

 intersection

1.23

 table

1.20

[*

1.17

[+]

1.15

 seams

1.14

 Omni

1.13

most

1.11

Activations Density 0.012%

instances of the words "top" and "bottom" related to position or placement

No Comments

No Known Activations