Isticmaal AI Open Source Walba Qaab Free Ah

Cashar gacan-ku-hays ah oo aan ku tusayo sida aad models open source ah kombiyuutarkaaga ugu run-gareyn karto, waana bilaash. Waxaan ugu talagalay developer-ka doonaya inuu AI ku dhex dhiso applications-kiisa isagoon API lacag ah bixin.

Waxaan ka billaabayaa Hugging Face, oo ah sida GitHub oo kale laakiin models-ka loogu talagalay. Waxaan tusayaa DeepSeek, LLaMA, Qwen, Gemma iyo Flux, iyo sida cabbirka model-ka uu ugu xirnaado RAM-ka kombiyuutarkaaga. Kadib waxaan soo dejinayaa Ollama, waxaanan terminal-ka ku run-gareynayaa DeepSeek-R1. Waxaad arkaysaa reasoning-ka: intuusan jawaabta ku siin ayuu ka fikirayaa oo is-weydiinayaa. Qaybta developers-ka waxaan sharraxayaa API-ga Ollama: endpoint-ka generate, POST request-ka, body-ga JSON-ka ah, model-ka iyo prompt-ka, iyo farqiga u dhexeeya streaming response iyo mid hal mar imanaya.

Waxaan sidoo kale ku xirayaa interface la yiraahdo Chatbox, waxaanan tusayaa habaynta environment variables-ka Windows-ka, OLLAMA_HOST iyo OLLAMA_ORIGINS, iyo tijaabinta API-ga Postman. Ugu dambayn waxaan sharraxayaa sida ay models uncensored-ka ah uga duwan yihiin kuwa kale, iyo in casharka xiga uu noqonayo Cursor.

A 29-minute Somali-language (Af-Soomaali) hands-on tutorial from Dugsiiye on running open-source AI models locally for free. It starts with Hugging Face as the GitHub of models, covering DeepSeek, LLaMA, Qwen, Gemma and Flux, and how model size maps to the RAM your machine needs. It then installs Ollama, runs DeepSeek-R1 in the terminal and shows the reasoning step before the answer. The developer half explains the Ollama API: the generate endpoint, POST requests, the JSON body with model and prompt, and the difference between a streaming response and a single complete one. It also connects Chatbox as a front end, sets OLLAMA_HOST and OLLAMA_ORIGINS on Windows, and tests the API in Postman. Free on YouTube.