INDEX

Explanations

references to corporate leadership and executive positions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 26-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L26R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.26.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

Associate

-0.16

 Associate

-0.15

 Kral

-0.14

ãĥ¼ãĥĦ

-0.14

maal

-0.14

ossal

-0.13

 znal

-0.13

 Huck

-0.13

LOCKS

-0.13

vented

-0.13

POSITIVE LOGITS

CEO

0.38

CEO

0.34

 chairman

0.30

 Chairman

0.30

ce

0.28

 president

0.28

CE

0.27

CO

0.24

 executive

0.24

Chair

0.24

Activations Density 0.088%

references to corporate leadership and executive positions

No Comments

No Known Activations