Elrich akira

elrich666's activity

• All
• Models
• Datasets
• Spaces
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

liked a Space about 1 month ago

Running on A10G

1.39k

1.39k

GGUF My Repo

🦙

Create and quantize Hugging Face models

upvoted a collection about 1 month ago

Muy denunciado = Debe ser muy bueno

Collection

Si es muy denunciado es muy probable de que sea un modelo bastante bueno. Si está en este listado es garantía de calidad. • 1 item • Updated Mar 8 • 1

liked a model about 2 months ago

Novaciano/Llama-3.2-3b-NSFW_Aesir_Uncensored-GGUF

Updated Feb 7 • 334 • 11

reacted to Jaward's post with 👍🔥 about 2 months ago

view post

Post

1528

The beauty in GRPO is the fact that it doesn’t care if the rewards are rule-based or learned, the hack: let the data self-normalize— trajectories in a batch compete against their mean, no value model, no extra params, just clean, efficient RL that cuts memory usage by 50%, while maintaining SOTA performance. btw it was introduced 9months prior to R1: arxiv.org/pdf/2402.03300

See translation

• 
• 
• 

• 
• 1 reply

·

🔥

7

7

👍

1

1

+

Reply