INDEX

Explanations

specific proper nouns and terms related to names and titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 9-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L9R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.9.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

uddle

-0.17

stile

-0.15

Vys

-0.14

mons

-0.14

 Grund

-0.14

met

-0.14

iare

-0.14

èĩªæĭį

-0.13

peat

-0.13

ingly

-0.13

POSITIVE LOGITS

mada

0.16

 pres

0.15

zza

0.13

 Pres

0.13

pty

0.13

leÅŁik

0.13

 Reign

0.12

 excer

0.12

Enumer

0.12

iyim

0.12

Activations Density 0.039%

specific proper nouns and terms related to names and titles

No Comments

No Known Activations