INDEX

Explanations

moments when content is informative or thought-provoking

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

cept

-0.15

uts

-0.14

å¦Ļ

-0.14

addir

-0.13

forgettable

-0.13

 kapat

-0.13

isans

-0.13

ores

-0.12

imento

-0.12

ilog

-0.12

POSITIVE LOGITS

 Feel

0.47

 feel

0.47

Feel

0.47

feel

0.42

 Enjoy

0.34

Enjoy

0.32

 enjoy

0.31

 please

0.30

 hope

0.30

 Please

0.30

Activations Density 0.336%

moments when content is informative or thought-provoking

No Comments

No Known Activations