INDEX

Explanations

phrases or structures that indicate conjunctions and complex sentence construction

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 24-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L24R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.24.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

 whereas

-0.16

iner

-0.14

 pÅĻiÄįemÅ¾

-0.14

islav

-0.13

 ÙĪØ§ÙĦØªÙĬ

-0.12

 because

-0.12

 entitled

-0.12

gaard

-0.12

 undergo

-0.12

agini

-0.12

POSITIVE LOGITS

ients

0.32

/or

0.26

 indeed

0.23

 especially

0.22

uxtap

0.19

its

0.19

 Ð¾ÑģÐ¾Ð±ÐµÐ½Ð½Ð¾

0.19

phans

0.19

Ð¾ÑģÐ¾Ð±

0.18

lack

0.18

Activations Density 0.312%

phrases or structures that indicate conjunctions and complex sentence construction

No Comments

No Known Activations