INDEX

Explanations

references to apologies and announcements related to personal or professional activities

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

avra

-0.16

iyan

-0.15

 addCriterion

-0.14

ÂŃi

-0.14

ÏĦÏħ

-0.14

 à¹Ĩ

-0.14

.failure

-0.14

challenge

-0.13

ymb

-0.13

ISMATCH

-0.13

POSITIVE LOGITS

 fans

0.26

 Fans

0.25

fan

0.23

Fans

0.22

fans

0.22

Fan

0.19

 group

0.19

 idols

0.19

fan

0.19

 fandom

0.18

Activations Density 0.001%

references to apologies and announcements related to personal or professional activities

No Comments

No Known Activations