INDEX

Explanations

comparative phrases or structures indicating preferences or evaluations

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_L5_Hcat_z_lr1.20e-03_l11.00e+00_ds49152_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

Data Type

float32

Hook Name

blocks.5.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

2:0.16

3:0.05

4:0.10

5:0.03

6:0.03

7:0.25

8:0.13

9:0.03

10:0.04

11:0.09

Negative Logits

��

-1.68

 equivalents

-1.68

aleb

-1.61

assad

-1.61

pez

-1.61

 Horses

-1.53

UGH

-1.48

�

-1.48

一

-1.47

¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

-1.44

POSITIVE LOGITS

 affair

1.50

mie

1.50

xious

1.43

 Strait

1.41

 reliant

1.35

 affecting

1.35

angered

1.33

leneck

1.30

aceae

1.29

TMZ

1.28

Activations Density 0.002%

comparative phrases or structures indicating preferences or evaluations

No Comments

No Known Activations