All my families of decoder-only models trained from scratch
🌌 Local LLM Efficiency
DedeProGames
DedeProGames
AI & ML interests
Agentic & Coding finetune, Decoder-Onlys from Scratch, Image LoRAs, AI Researcher
Recent Activity
liked a model about 1 hour ago
Qwen/Qwen3.5-2B updated a dataset 8 days ago
DedeProGames/lm-tetris-arena-results liked a Space 11 days ago
DedeProGames/Supra2-Medium-VL-ChatOrganizations
GPT-U: small Llama-style base models trained from scratch on web, educational text and code, with fully open scripts, logs and evals.
DynamicMind: a family of lightweight models trained from scratch on diverse datasets.
Some experimental decoder-onlys made by me for research
Experimental medical model based on Qwen2.5 Family
All my families of decoder-only models trained from scratch
Kiyo: Family of models trained on large-scale web, educational, and code data.
GPT-U: small Llama-style base models trained from scratch on web, educational text and code, with fully open scripts, logs and evals.
Wisp: small Llama-style base models trained from scratch on educational text and code.
DynamicMind: a family of lightweight models trained from scratch on diverse datasets.
LowOnMind: Small decoder-only models trained in small amount of tokens
Some experimental decoder-onlys made by me for research
Experimental medical model based on Qwen2.5 Family
Small and Efficient Chess Models