Software Developer • 2m
1. LoRA on a reasonably small open model (best balance for small compute): apply low-rank adapters (PEFT/LoRA). Requires less GPU memory and works well for 700–3000 rows. 2. Full fine-tune (costly / heavy): only if you have >A100 GPU or cloud paid GPU. Not recommended for early MVP. 3. No-fine-tune alternative (fast & free): use retrieval + prompting (RAG) — keep base LLM and add context from your 3k+ rows. Great when compute is limited.
Willing to contribut... • 1m
I fine-tuned 3 models this week to understand why people fail. Used LLaMA-2-7B, Mistral-7B, and Phi-2. Different datasets. Different methods (full tuning vs LoRA vs QLoRA). Here's what I learned that nobody talks about: 1. Data quality > Data quan
See More
AI Deep Explorer | f... • 7m
LLM Post-Training: A Deep Dive into Reasoning LLMs This survey paper provides an in-depth examination of post-training methodologies in Large Language Models (LLMs) focusing on improving reasoning capabilities. While LLMs achieve strong performance
See MoreAI Deep Explorer | f... • 8m
"A Survey on Post-Training of Large Language Models" This paper systematically categorizes post-training into five major paradigms: 1. Fine-Tuning 2. Alignment 3. Reasoning Enhancement 4. Efficiency Optimization 5. Integration & Adaptation 1️⃣ Fin
See More
Hey I am on Medial • 4m
The "AI Personality Forge: Product :"Cognate" Cognate is not an AI assistant. It's a platform on your phone or computer that lets you forge, train, and deploy dozens of tiny, specialized AI "personalities" for specific tasks in seconds. How it wo
See MoreDownload the medial app to read full posts, comements and news.