INDEX

Explanations

references to visits and experiences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 26-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L26R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.26.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

avra

-0.17

 unexpectedly

-0.17

igin

-0.16

usta

-0.15

-valu

-0.15

ackbar

-0.15

antino

-0.15

anlar

-0.14

InstanceState

-0.14

 realize

-0.14

POSITIVE LOGITS

 definitely

0.40

 Definitely

0.36

 certainly

0.31

 Certainly

0.29

 definite

0.27

 definit

0.23

 DEFIN

0.22

 Absolutely

0.22

 absolutely

0.21

 surely

0.21

Activations Density 0.029%

references to visits and experiences

No Comments

No Known Activations