INDEX

Explanations

phrases related to how-to instructions and steps for various tasks

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 20-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L20R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.20.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

 somehow

-0.18

.idea

-0.16

 somewhere

-0.16

borough

-0.15

edef

-0.15

ÑıÐ¶

-0.14

Ã©rÃ©

-0.14

 Reason

-0.14

anda

-0.14

ILON

-0.14

POSITIVE LOGITS

 yourself

0.28

 effectively

0.24

 Yourself

0.22

 oneself

0.22

 your

0.20

your

0.20

 effective

0.19

 yourselves

0.19

æľīæķĪ

0.18

ä½łçļĦ

0.18

Activations Density 0.289%

phrases related to how-to instructions and steps for various tasks

No Comments

No Known Activations