INDEX

Explanations

terms related to visitors or participants in events or 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

 becoming

-0.15

.listFiles

-0.15

Î³Î¼Î±

-0.14

antor

-0.14

anax

-0.14

atern

-0.14

ÑģÑĤÐµÑĢ

-0.13

 Ð¿ÑĢÐ¸Ð·

-0.13

.Xaml

-0.13

Ģë¡ľ

-0.13

POSITIVE LOGITS

 should

0.27

who

0.27

can

0.25

who

0.23

 interested

0.23

 shouldn

0.22

may

0.22

 will

0.20

 wishing

0.19

interested

0.18

Activations Density 0.120%

terms related to visitors or participants in events or activities

No Comments

No Known Activations