INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Configuration

ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L7_Hcat_z_lr1.20e-03_l11.10e+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.7.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

0:0.07

1:0.08

2:0.08

3:0.09

4:0.08

5:0.07

6:0.08

7:0.08

8:0.08

9:0.08

10:0.08

11:0.07

Negative Logits

 dressing

-2.61

 winter

-2.38

mas

-2.26

 marriage

-2.23

 weddings

-2.12

 inflation

-2.07

bow

-2.06

 seasons

-2.06

 season

-2.05

 anthem

-2.04

POSITIVE LOGITS

 Takeru

2.30

]=

2.27

plet

2.25

 Tsarnaev

2.20

cade

2.13

anooga

2.08

seless

2.06

ギ

2.05

yg

2.04

skill

2.01

Activations Density 0.000%

No Known Activations

This feature has no known activations.