QwenLM/ParScale
Parallel Scaling Law for Language Model — Beyond Parameter and Inference Time Scaling
ParScale introduces a new way to make large language models (LLMs) more capable without drastically increasing their size or slowing them down. By running multiple variations of an input through the model in parallel, it generates richer, more accurate outputs, particularly for complex tasks like coding or math. This is designed for AI researchers and engineers who build and deploy LLMs.
476 stars. No commits in the last 6 months.
Use this if you need to enhance the performance and reasoning abilities of your LLMs, especially for tasks requiring deep understanding, while being mindful of computational resources like memory and inference time.
Not ideal if your primary goal is to simply scale model parameters or inference speed using traditional methods, as this introduces a new parallel computation paradigm.
Stars
476
Forks
24
Language
Python
License
—
Category
Last pushed
May 17, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/QwenLM/ParScale"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
jncraton/languagemodels
Explore large language models in 512MB of RAM
microsoft/unilm
Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities
haizelabs/verdict
Inference-time scaling for LLMs-as-a-judge.
albertan017/LLM4Decompile
Reverse Engineering: Decompiling Binary Code with Large Language Models
bytedance/Sa2VA
Official Repo For Pixel-LLM Codebase