INDEX

Explanations

This neuron is looking for specific terms related to stories or events in science fiction or fantasy contexts

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

words related to the concept of sabotage

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-att-kk

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

Not in Any Lists

No Comments

Head Attr Weights

0:0.05

1:0.02

2:0.23

3:0.06

4:0.24

5:0.05

6:0.03

7:0.04

8:0.05

9:0.09

10:0.05

11:0.02

Negative Logits

bage

-1.55

wikipedia

-1.49

nces

-1.46

steen

-1.42

miah

-1.34

ndra

-1.33

aceae

-1.28

MAP

-1.27

untarily

-1.26

lishes

-1.25

POSITIVE LOGITS

WARE

1.37

 bells

1.36

 pains

1.20

ooth

1.18

 labor

1.12

 ratt

1.06

oman

1.06

ioxide

1.05

Labor

1.05

deb

1.04

Activations Density 0.001%

This neuron is looking for specific terms related to stories or events in science fiction or fantasy contexts

words related to the concept of sabotage

No Comments

No Known Activations