INDEX

Explanations

phrases related to architectural features and structural elements

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

le

-0.21

(le

-0.17

 especially

-0.17

soever

-0.17

ÙĪØ§Ø¡

-0.16

Le

-0.16

 outs

-0.15

 especialmente

-0.15

el

-0.15

POSITIVE LOGITS

venue

0.30

la

0.30

ven

0.28

venues

0.24

 facto

0.22

son

0.21

venir

0.21

stru

0.20

velope

0.20

 cette

0.19

Activations Density 0.023%

phrases related to architectural features and structural elements

No Comments

No Known Activations