INDEX

Explanations

references to the "Halo" video game series and its specific titles and elements

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

0:0.03

1:0.03

2:0.08

3:0.16

4:0.02

5:0.03

6:0.18

7:0.16

8:0.04

9:0.09

10:0.08

11:0.05

Negative Logits

OHN

-1.74

ァ

-1.23

emen

-1.17

�

-1.16

tle

-1.12

OY

-1.10

ェ

-1.09

 Seymour

-1.09

 Hond

-1.09

OAD

-1.05

POSITIVE LOGITS

��

1.21

 Credits

1.15

 exile

1.10

 Tempest

1.09

athlon

1.04

 distances

1.03

etus

1.02

bash

1.00

 Deliver

0.99

monary

0.97

Activations Density 0.006%

references to the "Halo" video game series and its specific titles and elements

No Comments

No Known Activations