INDEX

Explanations

technical terms related to science, research, and technology

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

Head Attr Weights

0:0.09

1:0.14

2:0.04

3:0.06

4:0.06

5:0.07

6:0.10

7:0.04

8:0.06

9:0.14

10:0.09

11:0.04

Negative Logits

 sucker

-1.65

aterasu

-1.51

example

-1.45

 motif

-1.44

perfect

-1.40

akespeare

-1.38

 Tolkien

-1.37

 quint

-1.36

oliath

-1.35

advant

-1.34

POSITIVE LOGITS

News

2.04

Report

1.94

 Updates

1.81

NEWS

1.80

ews

1.75

Reports

1.70

Alert

1.62

 watchdog

1.57

Email

1.57

 Emails

1.56

Activations Density 0.047%

technical terms related to science, research, and technology

No Known Activations