Memuat…

Course Forum Asisten Sumber Cara belajar RAG Peta modul

Masuk Mulai →

socrates

Semua tentang AI, satu jalur belajar. Gratis, bersumber terbuka, dan setiap jawaban punya rujukan.

Navigasi

Course
Forum
Asisten
Sumber
Cara belajar
RAG
Peta modul

Dibangun dengan

Next.js · Tailwind ·
Lenis + GSAP · SQLite + BM25
(RAG hybrid: cosine + BM25, RRF)

© 2026 socratesgratis · terbuka · sumber selalu dicantumkan

I3 · Distributed Serving & Parallelism Strategies — socrates · socrates

LLM Infrastructure/Lintasan Inti/I3

I3

Distributed Serving & Parallelism Strategies

Tensor Parallelism, Pipeline Parallelism, Expert Parallelism (MoE), Interconnect (NVLink, PCIe)

Durasi

1.5 mgg

Lesson

5

Project

1

Mode

2 jenis

Lesson

0/5 selesai

01
Kapan dan Mengapa Butuh Distributed GPU Serving?
02
Tensor Parallelism (TP): Megatron-LM Style
03
Pipeline Parallelism (PP): Pembagian Layer Lintas Device
04
Mixture-of-Experts (MoE) & Expert Parallelism (EP)
05
Project: Multi-GPU Cluster Simulation & TP Profiling

← I2Inference Engines: vLLM, PagedAttention & Continuous Batching I4 →Advanced Optimization: Quantization, Speculative Decoding & Production LLMOps