INDEX

Explanations

references to lost or missing historical records or archives

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 14-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L14R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.14.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

Ø¬ÛĮ

-0.16

ocha

-0.16

hei

-0.14

Î¬Î»Î¹

-0.14

 Ð´Ð½Ñı

-0.14

 deepest

-0.14

æľĢæĸ°

-0.14

rar

-0.14

flt

-0.14

à¸£à¸Ķ

-0.14

POSITIVE LOGITS

 lost

0.22

lost

0.20

 Lost

0.18

 perish

0.18

 losses

0.18

ext

0.17

å½ĵ

0.17

ersist

0.16

 missing

0.16

 Gone

0.16

Activations Density 0.179%

references to lost or missing historical records or archives

No Comments

No Known Activations