INDEX

Explanations

references to high-engagement or popular topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 4-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L4R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.4.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

aci

-0.18

utor

-0.16

ickle

-0.15

 dwarf

-0.14

audi

-0.14

.mapping

-0.14

cket

-0.14

naments

-0.14

otent

-0.14

enez

-0.14

POSITIVE LOGITS

 stake

0.15

rax

0.15

 apprec

0.14

 spot

0.14

 welded

0.14

ITE

0.14

urname

0.14

statt

0.13

cape

0.13

 Wool

0.13

Activations Density 0.001%

references to high-engagement or popular topics

No Comments

No Known Activations