INDEX

Explanations

food-related terms and ingredients

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

Head Attr Weights

0:0.04

1:0.06

2:0.03

3:0.04

4:0.02

5:0.44

6:0.03

7:0.02

8:0.09

9:0.10

10:0.06

11:0.02

Negative Logits

quickShipAvailable

-1.93

iT

-1.58

 unaff

-1.54

�醒

-1.51

bart

-1.49

 upfront

-1.45

Requ

-1.44

LW

-1.43

 uncomp

-1.42

 Sund

-1.41

POSITIVE LOGITS

uminati

1.90

chu

1.68

cha

1.56

 Minion

1.55

uster

1.50

uchi

1.50

lopp

1.47

ventures

1.46

veyard

1.45

igans

1.43

Activations Density 0.400%

food-related terms and ingredients

No Known Activations