INDEX

Explanations

the word "you" in various contexts, indicating a focus on direct address or engagement with the reader

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

utor

-0.17

raya

-0.15

cem

-0.14

llum

-0.14

ahren

-0.14

ÙĬÙĥÙĬ

-0.14

poss

-0.13

anela

-0.13

orem

-0.13

ray

-0.13

POSITIVE LOGITS

 probably

0.29

 recall

0.25

 hopefully

0.23

 noticed

0.22

may

0.21

 well

0.20

Probably

0.20

 already

0.20

 Probably

0.20

 notice

0.20

Activations Density 0.047%

the word "you" in various contexts, indicating a focus on direct address or engagement with the reader

No Comments

No Known Activations