INDEX

Explanations

negations and conditional statements

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

0:0.02

1:0.04

2:0.14

3:0.16

4:0.02

5:0.04

6:0.09

7:0.10

8:0.07

9:0.12

10:0.12

11:0.04

Negative Logits

Motor

-1.14

dry

-1.08

sale

-1.08

Ale

-1.07

Nev

-1.07

XXX

-1.05

ーティ

-1.05

-1.04

went

-1.03

Private

-1.03

POSITIVE LOGITS

ramid

1.27

roup

1.24

izoph

1.17

 TAMADRA

1.11

 Tsukuyomi

1.11

 gobl

1.09

 interven

1.06

 cataly

1.05

 disag

1.03

 uptake

1.01

Activations Density 0.005%

negations and conditional statements

No Comments

No Known Activations