INDEX

Explanations

the use of the phrase "first up" and its variations, indicating an introductory sequence or listing

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 2-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L2R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.2.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

YRO

-0.16

nud

-0.15

ollen

-0.15

Ø¹Ø§Øª

-0.14

à¹Ģà¸ī

-0.14

ulk

-0.14

 shower

-0.14

_nth

-0.14

rico

-0.13

å£«

-0.13

POSITIVE LOGITS

isd

0.17

Ã¼zel

0.15

ruz

0.15

 responders

0.15

ucci

0.14

oston

0.14

mb

0.14

 Ø§ÙĦÙĤØ¯

0.14

 wors

0.13

Activations Density 0.103%

the use of the phrase "first up" and its variations, indicating an introductory sequence or listing

No Comments

No Known Activations