INDEX

Explanations

names and titles related to professional roles and conferences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 24-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L24R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.24.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

-chan

-0.17

æ£

-0.15

avl

-0.14

hiba

-0.14

abeth

-0.13

icut

-0.13

outh

-0.13

brtc

-0.13

Ð¾

-0.13

anol

-0.13

POSITIVE LOGITS

 French

0.23

 France

0.23

France

0.21

 french

0.21

French

0.20

 france

0.19

æ³ķåĽ½

0.19

 Lyon

0.18

.fr

0.18

 paris

0.18

Activations Density 0.066%

names and titles related to professional roles and conferences

No Comments

No Known Activations