MakazhanAlpamys

微调大模型，配置写到头大？ Soup 一个 YAML 就能开训，层流式加载让 8B 模型在 4GB 显存笔记本上跑，一条命令搞定，不用 SSH、不用折腾配置。 适合显存小、想自己微调的人，先从小模型试，别直接上大参数。 你微调先练什么，聊天还是分类？ #微调 #大模型 @grok MakazhanAlpamys/Soup

Top 5 AI GitHub Repositories That Feel Illegal to Know (100% Free & Local) Stop wasting $200/month on proprietary AI subscriptions. In this video, we break down the Top 5 Game-Changing Open Source AI GitHub Repositories you can run 100% free on your local PC — replacing ChatGPT Plus, Claude Pro, expensive scraper APIs, and cloud GPU rentals forever! ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ Your finding this on YouTube : ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1. This GitHub Repo Is Full Of Free API’s (All Categories) 2. mattpocock/skills: A complete AI Coding workflow, end-to-end 3. The Only GitHub Guide You’ll Ever Need 4. I Tried 100+ Claude Code Skills. These 6 Are The Best 5. My Opencode Workflow As A Senior Engineer 6. How To Make Your GitHub Stand Out (Gets You Hired!) 7. Ab lazy hokr bhi paise bna skte ho XD. Github Repo - Money Printer Turbo 8. 4 Claude Code GitHub Repo 9. AI-Assisted Coding Tutorial – OpenClaw, GitHub Copilot, Claude Code, CodeRabbit, Gemini CLI 10. [Free on Github] My Jarvis AI Assistant 11. 5 Github repos that print $$$ 12. 7 New Github Repos You Need NOW (GPT-6 Astra) 13. 7 Free GitHub Repos That Make Claude So Good It Feels Illegal 14. Top Repos + Fame, Traffic & Agents 15. 5 Github Repos for Claude = 3 lakh/month 16. Insane GitHub Repos That 10x Your Codex And Claude Code Setup 17. Free on GitHub: manage agents, 11Labs alternative, incredible design & more 18. [FREE on GitHub] I Built My Own Real Jarvis Interface 19. 9 shocking repos that make money 20. Top Repos Explained: Code Review, World Camera, Better AI Builds, etc. ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ ⏱️ TIMESTAMPS / CHAPTERS ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 00:00 - The $200/Mo AI Trap & Open Source Revolution 00:19 - #1 Firecrawl: Web Scraper for LLM Agents (190k★) 00:42 - [COMMUNITY] Join Telegram & Get Free Setup Scripts 00:46 - #2 AnythingLLM: Private Local AI Workspace (67k★) 01:08 - #3 Agent-Reach: Zero-Fee Live Web Search (95k★) 01:35 - [RESOURCES] Free Docker & Config Templates 01:38 - #4 Screenpipe: 24/7 Local Memory Engine (22k★) 02:00 - #5 Soup: Fine-Tune 8B on 4GB Laptop GPU 02:25 - [ROADMAP] Build Your Own AI Stack (Outro) ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 🛠️ FEATURED GITHUB REPOSITORIES ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1. Firecrawl (mendableai/firecrawl) ★ 190.4k Stars | AGPL-3.0 | Turn messy websites into clean LLM markdown Repo: firecrawl/firecrawl Quickstart: git clone firecrawl/firecrawl && docker compose up -d 2. AnythingLLM (Mintplex-Labs/anything-llm) ★ 67.0k Stars | MIT | Private desktop/server local ChatGPT + RAG workspace Repo: Mintplex-Labs/anything-llm Quickstart: docker run -d -p 3001:3001 -v anythingllm:/app/server/storage mintplexlabs/anythingllm 3. Agent-Reach (Panniantong/Agent-Reach) ★ 95.8k Stars | MIT | Free live web search for AI agents without API keys Repo: Panniantong/Agent-Reach Quickstart: pip install agent-reach && agent-reach search "latest AI papers" 4. Screenpipe (mediar-ai/screenpipe) ★ 21.9k Stars | Apache 2.0 | 24/7 local screen OCR & audio episodic memory Repo: mediar-ai/screenpipe Quickstart: cargo install screenpipe && screenpipe 5. Soup (MakazhanAlpamys/Soup) ★ 8.5k Stars | Apache 2.0 | Layer streaming trains 8B models on 4GB laptop GPU Repo: MakazhanAlpamys/Soup Quickstart: pip install trysoup && soup train config.yaml ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 📥 DOWNLOAD FREE SETUP GUIDES & PROMPTS (OFFICIAL REPO) ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 📦 Master Architecture Guide (All 5 Repos in 1 Complete File): 👉 Direct Link: t.me 🔍 In Telegram Search: "MASTER ARCHITECTURE GUIDE" 📄 Modular Standalone Setup Guides (Single Repo Downloads): 1️⃣ Firecrawl Deep Setup: 👉 Link: t.me | 🔍 Search: "Guide 01: Firecrawl" 2️⃣ AnythingLLM Private Local RAG: 👉 Link: t.me | 🔍 Search: "Guide 02: AnythingLLM" 3️⃣ Agent-Reach Autonomous Search: 👉 Link: t.me | 🔍 Search: "Guide 03: Agent-Reach" 4️⃣ Screenpipe 24/7 AI Memory: 👉 Link: t.me | 🔍 Search: "Guide 04: Screenpipe" 5️⃣ Soup AI 4GB GPU Fine-Tuning: 👉 Link: t.me | 🔍 Search: "Guide 05: Soup AI" ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 🌐 OFFICIAL TECH GVS COMMUNITY LINKS ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 📢 Telegram (First Access & Code): t.me 💻 Official Website: techgvs.in 🎥 YouTube Channel: youtube.com 💬 WhatsApp Channel: whatsapp.com ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ Your Queries : ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ agent reach ai web search fine tune 8b model on 4gb gpu soup ai fine tuning guide ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ #LocalAI #AnythingLLM #Firecrawl #TechGVS #Python #SelfHosted #AIWorkflows #LLM © 2026 Tech GVS | Website: techgvs.in

Top 5 open source do GitHub | Vol. 10 #shorts Os projetos #5 a #1 do Top 15 open source do GitHub, Vol. 10, 10 segundos para cada um. #5 MailCheap: marclou/mailcheap #4 Laranode: crivion/laranode #3 OneCLI: onecli/onecli #2 Soup: MakazhanAlpamys/Soup #1 RemoveMacAI: omlahore/RemoveMacAI Vídeo completo com os 15: youtu.be Oferecimento: Hallen Tech, hallentech.com #shorts #opensource #github

15 projetos open source do GitHub que valem o seu tempo | Vol. 10 15 projetos open source do GitHub que valem o seu tempo, em contagem regressiva e com 10 segundos para cada um: de estúdio de vídeo para o seu agente de código a painel de hospedagem, GIS em Rust e uma faxina no Apple Intelligence do Mac. 📺 Décimo volume da série Open source do GitHub, o primeiro com 15 projetos. Escolhidos numa nova análise da lista inteira: estrelas, frequência de commits, último commit, README, documentação, capturas de tela e site, conferidos em 20/10/2026. Qual desses você vai testar primeiro? Conta nos comentários! ⏱️ CAPÍTULOS E LINKS 0:00 Abertura 0:10 #15 showtime: estúdio de vídeo local para o seu agente de código, sem nuvem FavioVazquez/showtime 0:20 #14 Worklenz: gestão de projetos open source com lista, quadro e Gantt, na nuvem ou no seu servidor (3,3 mil estrelas) Worklenz/worklenz 0:30 #13 Reminal: feche a tampa do Mac e use as janelas, os desktops e os terminais dele em qualquer navegador harshalgajjar/Reminal 0:40 #12 M3E Canvas: desenhe telas Material 3 Expressive no navegador e copie um prompt para a sua IA de código (9,5 mil estrelas) lnkiai/m3e-canvas 0:50 #11 TapMap: mapa em tempo real de para onde o seu computador se conecta, local e sem telemetria (1,6 mil estrelas) olalie/tapmap 1:00 #10 Reitti: linha do tempo da sua localização no seu servidor, com compartilhamento ao vivo (2,7 mil estrelas) dedicatedcode/reitti 1:10 #9 Television: buscador fuzzy rapidíssimo para o terminal (6,4 mil estrelas) alexpasmantier/television 1:20 #8 RustGIS: GIS gratuito em Rust, no Windows, macOS, Linux e no navegador (1,7 mil estrelas) opengeos/RustGIS 1:30 #7 Langfuse: observabilidade, prompts e avaliações para aplicações de IA, com self-host (36,1 mil estrelas) langfuse/langfuse 1:40 #6 Hindsight: memória para agentes de IA que aprendem com o tempo (51,7 mil estrelas) vectorize-io/hindsight 1:50 #5 MailCheap: newsletter self-hosted em que você paga pelos e-mails enviados, com MCP para o seu agente (1,3 mil estrelas) marclou/mailcheap 2:00 #4 Laranode: painel de hospedagem open source, alternativa ao cPanel e ao Plesk (1,5 mil estrelas) crivion/laranode 2:10 #3 OneCLI: um agente de IA isolado para cada pessoa da equipe, com cofre de credenciais e regras de acesso (3,6 mil estrelas) onecli/onecli 2:20 #2 Soup: fine-tuning de LLMs com um comando, até modelo de 8B em GPU de 4 GB (modo beta) (9,4 mil estrelas) MakazhanAlpamys/Soup 2:30 Oferecimento Hallen Tech 2:44 #1 RemoveMacAI: desliga o Apple Intelligence, apaga os modelos e corta anúncios e coleta de dados do macOS, tudo reversível (5 mil estrelas) omlahore/RemoveMacAI 2:54 Encerramento 🧡 Oferecimento: Hallen Tech, liderança de tecnologia sob demanda, software, cloud e segurança para a sua empresa. Site: hallentech.com E-mail: contato@hallentech.com Instagram: instagram.com LinkedIn da Hallen Tech: linkedin.com LinkedIn do Eddwan: linkedin.com Se curtiu, deixa o like, segue o canal e ativa as notificações para a próxima lista. #opensource #github #programacao #ferramentas

repo 👇 MakazhanAlpamys/Soup layer streaming is still beta and opt-in. the README has a free Colab notebook so you can check the 4 GB claim yourself

@mouad_builds I used Qwen3-8B before and fine-tuned the tool-calling model using this open-source MakazhanAlpamys/Soup.

MakazhanAlpamys/Soup You can now train an AI model on a 4GB laptop running a 3050. Incredible work.

🔗 영상에서 소개한 오픈소스 · vectorize-io/hindsight — vectorize-io/hindsight 👀 오늘 함께 뜬 리포 · morluto/rea — morluto/rea · MakazhanAlpamys/Soup — MakazhanAlpamys/Soup 매일 개발자가 실제로 Star 누른 오픈소스를 하나씩 뜯어봅니다. 구독하면 놓치지 않아요. 🗂 오늘 분석한 10개 전부와 지난 기록: onerepoaday.onerepoaday.workers.dev 📬 매주 일요일, 이번 주 가장 뜬 AI 오픈소스 Top 10을 메일 한 통으로 보내드립니다 page.stibee.com

컴퓨터는 왜 소수 계산을 틀릴까? 원리 한 개: 부동소수점 1. 소수를 이진 동전으로 2. 끝을 잘라 반올림 3. 순서와 비교가 흔들림 오늘 깃허브에서 뜬 리포: vectorize-io/hindsight vectorize-io/hindsight 오늘의 다른 상위 픽: · morluto/rea — morluto/rea · MakazhanAlpamys/Soup — MakazhanAlpamys/Soup 🗂 오늘 분석한 10개 전부와 지난 기록: onerepoaday.onerepoaday.workers.dev 📬 매주 일요일, 이번 주 가장 뜬 AI 오픈소스 Top 10을 메일 한 통으로 보내드립니다 page.stibee.com 뉴스가 아니라 개발자의 행동(Star)을 추적합니다. 매일 새벽 5시 자동 업데이트. #Shorts #AI #오픈소스 #GitHub

Soup: Fine-tune and post-train LLMs in one command. No SSH, no config hell. (9.1k stars) What it does All text below is quoted directly from the project README. >Fine-tune and post-train LLMs in one command. No SSH, no config hell. >Soup turns the pain of LLM fine-tuning into a simple workflow. One config, one command, done. >Fine-tune an 8B model on a 4 GB laptop GPU. Layer streaming keeps the frozen base out of VRAM and feeds it to the GPU one decoder layer at a time. Measured on an RTX 3050 Laptop 4 GB: Llama-3.1-8B-Instruct + NF4 at 119.6 tok/s at 952 MHz, 3.32 GB peak (pre-repair), bit-exact against a normal resident run, and reproduced independently on an H100 at 113.00 tok/s in the same 3.32 GB. >Opt-in (stream_layers: true) and still BETA. Why Soup? >Training LLMs is still painful. Even experienced teams spend 30-50% of their time fighting infrastructure instead of improving models. Soup fixes that. - Zero SSH. Never SSH into a broken GPU box again. - One config. A simple YAML file is all you need. - Auto everything. Batch size, GPU detection, quantization: handled. - Works locally. Train on your own GPU with QLoRA. No cloud required. What's New >v0.76.0: settings Soup accepted and never applied now stop the load. Options that were validated and documented, and read by nothing on the path they claimed to configure, are now refused at load, naming the setting and what to change; the rest warn and name the release that will refuse them. >Validation loss is now evaluated. data.val_split held rows out of training on every run, and no trainer on the transformers backend evaluated them. Every task that gets a split now evaluates it at the end of each epoch, or every N steps with training.eval_steps, and records the val loss; runs take a few percent longer. >Python 3.10-3.12 only. On 3.13+, pip used to resolve untested PyTorch wheels that crash in the native extension before Soup runs at all. >Soup is a command-line application, so the cleanest install gives it its own environment and puts soup on your PATH. >soup init, soup data ..., and the other data/inspection commands work on the light install. Fine-tuning (soup train) needs the [train] extra. Repository MakazhanAlpamys/Soup Repository stats |Metric|Value| |:-|:-| |Stars|9,066| |Forks|1,476| |Last release|v0.76.1 (2026-10-14)| |Languages|Python 98.7%, Jupyter Notebook 1.1%| |Contributors|111| |AI contributor accounts (Claude / Codex / Cursor / Copilot)|claude (520 commits)| |AI co-authored commits|Claude: 522| Main contributor |Field|Value| |:-|:-| |Name|Alpamys Makazhan| |GitHub|@MakazhanAlpamys| |Commits to this repo|1,175 (top contributor)| |Bio|building startup| |Company / Location|@SoupTeam / San Francisco, CA| |X / Twitter|@JustAlpamys| |Followers|924| |Public repos|100| |On GitHub since|2024-11-13|