INDEX

Explanations

specific types of character names or titles with notable connections to their contexts

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

soever

-0.15

umor

-0.14

'gc

-0.14

createQuery

-0.14

itating

-0.14

idunt

-0.14

isode

-0.14

mainwindow

-0.13

 iParam

-0.13

#:

-0.13

POSITIVE LOGITS

sonian

0.19

zheimer

0.18

bes

0.18

ocene

0.17

esian

0.17

-shaped

0.17

apanese

0.16

uvian

0.16

Qed

0.15

avian

0.15

Activations Density 0.415%

specific types of character names or titles with notable connections to their contexts

No Comments

No Known Activations