INDEX

Explanations

references to architectural elements and design details

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 30-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L30R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.30.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

IBC

-0.14

Ã¢ng

-0.14

atform

-0.14

Kit

-0.14

.EventQueue

-0.14

ouro

-0.14

 McInt

-0.14

ASURE

-0.14

 Dodd

-0.14

lobal

-0.13

POSITIVE LOGITS

 Paris

0.24

 paris

0.23

 Marseille

0.23

rue

0.22

 France

0.21

 Gaul

0.19

Paris

0.19

 Lyon

0.19

 Saint

0.18

 Louis

0.17

Activations Density 0.055%

references to architectural elements and design details

No Comments

No Known Activations