INDEX

Explanations

references to parallel structures or concepts within various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 21-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L21R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.21.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

Ø³Ùĩ

-0.15

essler

-0.15

zcze

-0.15

ela

-0.15

dess

-0.15

dur

-0.15

andes

-0.15

owns

-0.14

upil

-0.14

uet

-0.14

POSITIVE LOGITS

ism

0.34

izable

0.32

otope

0.23

lep

0.22

izers

0.22

ISM

0.21

 universe

0.21

izing

0.21

izer

0.21

ized

0.21

Activations Density 0.012%

references to parallel structures or concepts within various contexts

No Comments

No Known Activations