INDEX

Explanations

instances of rankings or ratings

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

er

-0.17

inic

-0.17

ucci

-0.16

eenth

-0.16

top

-0.16

zon

-0.15

specialchars

-0.15

esto

-0.15

Æ°á»Ľc

-0.15

POSITIVE LOGITS

pling

0.38

-notch

0.35

ographical

0.34

ography

0.31

most

0.30

 notch

0.29

amax

0.28

azes

0.28

az

0.27

-secret

0.27

Activations Density 0.053%

instances of rankings or ratings

No Comments

No Known Activations