AccelerateAI Careers

🚀 Hugging Face Introduces a Free Course on Fine-Tuning Local LLMs

Hugging Face is enhancing AI education with the launch of the Smöl Course, a free program designed to empower individuals in building, training, and fine-tuning machine learning models.

💡 Key Highlights:

👉 Practical Learning: Engage with real datasets and open-source tools for hands-on experience.
👉 Community Engagement: Collaborate and grow within an active AI community.
👉 Accessibility: Tailored for both beginners and experienced developers—no advanced degrees required.

💻 This course, available on GitHub, emphasizes smaller, efficient language models like SmolLM2, making AI more approachable and impactful for practical applications.

🌟 Course Topics Include:
👉 Instruction Tuning: Supervised fine-tuning, chat templating, and basic instruction following.
👉 Preference Alignment: Techniques like DPO and ORPO for aligning models with human preferences.
👉 Parameter-efficient Fine-tuning: Methods such as LoRA and prompt tuning.
👉 Evaluation: Automatic benchmarks and custom domain evaluations.
👉 Vision-language Models: Adapting multimodal models for vision-language tasks.
👉 Synthetic Datasets: Creating and validating synthetic datasets for training.
👉 Inference: Efficient model inference techniques.
👉 Deployment: Strategies for deploying and serving models at scale.

Small models can have a significant impact. Whether you're exploring AI or looking to enhance your skills, this course is designed for you.

👉 Start your journey today: github.com/huggingface/smol-course

Let’s work together towards a more inclusive AI future!

#ai #artificiallearning #ml #llm #generativeai #genai #huggingface #ailearning #aiandml #largelanguagemodels #nlp

1 year ago | [YT] | 3

AccelerateAI Careers

🧐 RAG versus Agentic RAG:

👇 Dive deep into the world of Retrieval-Augmented Generation (RAG) and discover how Agentic RAG is revolutionizing information processing. Understand the technical differences, architectural innovations, and real-world applications in this comprehensive breakdown.

🗝 Link to the full article: www.linkedin.com/pulse/rag-vs-agentic-snigdha-kakk…

#ai #accelerateaicareers #artificiallearning #rag #agenticrag #aiagents #retrieveraugmentedgeneration #techtrends #genai #generativeai

1 year ago | [YT] | 3

AccelerateAI Careers

🤔 Are you someone who is venturing newly into the field of AI and Popular Large Language Models??

🤓 Do you want to deep dive into Llama3 and know all about it at one place??

🧐 Then, this is for YOU!!!

👍 Subscribe 🛎 to my YouTube Channel ‪@AccelerateAICareers‬ I have a Playlist on Popular LLMs where I take Deep Dive into Architecture, Training Process, Testing Phase, Performance Benchmarks and Adoption of these LLMs. I also have another playlist on Generative AI and NLP Fundamentals where I am teaching a non-technical person how to become a data scientist in 2024.

#AI #GenerativeAI #artificialintelligence #career #careerswitch #datascience #datascientist #engineer #python #technology #machinelearning #dataanalyst #dataanalytics #careertransition #largelanguagemodels #llm #llms #llama3 #chatgpt #chatGPT4 #student #students #jobsincanada #canadalife #canadajobs

1 year ago (edited) | [YT] | 5

AccelerateAI Careers

⚡️ @Cohere open sources their LLM Chat interface and Cohere toolkit on GitHub⚡️

🔥This will be a major accelerant for getting LLMs into production and a great step towards Accessible AI 🤖

#accessibleai #accelerateaicareers #ai #artificialintelligence #machinelearning #ml #deeplearning #datascience #datascientists #ai #genai #generativeai #generative #llm #llms #largelanguagemodel #largelanguagemodels #cohere

1 year ago (edited) | [YT] | 4

AccelerateAI Careers

🔥Graph Machine Learning in the Era of LLMs

⚡️Provides a comprehensive overview of the latest advancements for Graph ML in the era of LLMs.

👇Covers the recent developments in Graph ML, how LLM can enhance graph features, and how it can address issues such as OOD and graph heterogeneity.

🧐According to the literature, LLMs can help advance Graph ML but there is also the potential of using graphs to enhance LLMs on both pre-training and inference.

What do you think 🤔 would be the future of Graph ML?? Mention in the comments!!

If you are looking for interesting research topics, SUBSCRIBE 🌟to my newsletter AI Scoop on LinkedIn and YouTube channel AccelerateAICareers

#ai #ml #deeplearning #data #datascience #datascientist #artificialintelligence #machinelearning #deeplearning #student #students #generative #generativeai #genai #llm #largelanguagemodel #largelanguagemodels #dataanalysis #dataengineer #dataengineering #softwaredeveloper #software

1 year ago (edited) | [YT] | 4

AccelerateAI Careers

👉The integration of AI in education has sparked a debate on its potential benefits and drawbacks.

👏 Proponents, such as those at EdSurge, argue that AI can recalibrate pedagogy and foster higher-order thinking.

🙄Others including educators and policymakers, express concerns about cheating and the need to maintain essential skills like critical thinking.

👉There is a great emphasis on the importance of navigating these challenges to ensure a balanced and effective integration of AI in the learning environment.

Here, in this article 📃 , it is presented how Google Gemini can transform education and learning for users!!
What do you think about it?? 🤔

#ai #ailearning #education #aieducation #DeepLearning #MachineLearning #artificalintelligence #Gemini #Google #aiteaching #GenerativeAI #generative #GenAI #largelanguagemodels #LLMs #foundationmodels #student #students #studentlife #engineer #engineering #developer #python #dataanalyst #softwaredeveloper #technology #tech

www.techlearning.com/news/google-gemini-how-google…

1 year ago | [YT] | 4

AccelerateAI Careers

😱 Phi-3 Mini (3.8 Billion) - #Microsoft introduces new 3.8B model, can fit on your iphone

📲 and still beats hashtag#Mixtral on MMLU 🤭........
here's what you need to know👇👇👇
Following in the footsteps of previous Phi models Microsoft has trained 3 new versions of the Phi #LLM. Their approach bucks the hashtag#Llama3 and #Mixtral trend of massive datasets, instead using smaller heavily filtered web data and synthetic data generated by larger models. The model is then trained in two phases,

1️⃣ initial exposure to general knowledge and language patterns, followed by
2️⃣ intensive training on logical reasoning and specific skills.
📱 Phi-3, available in 3 model sizes
📱 mini (3.8B), small (7B), and medium (14B)
📱 mini trained on 3.3T tokens
📱 small and medium trained on 4.8T
📱 Context lengths between 4K to 8K (extendable with RoPE)
📱 Performs competitively against Mixtral 8x7B and hashtag#GPT-3.5
📱 MMLU: 69% for mini, 75% for small, 78% for medium
📱 MT-bench: 8.38 for mini, 8.7 for small, 8.9 for medium
📱 Used DPO for safety alignment
📱 MIT Licence

🥳 Paper 👉 arxiv.org/pdf/2404.14219.pdf
Models 👉 huggingface.co/models?other=phi3&sort=trending&sea…

🟠 Want to learn more about LLMs? ....Check out our YouTube Playlist 👉 youtube.com/@AccelerateAICareers?sub_confirmation=… 🟠

1 year ago (edited) | [YT] | 4

AccelerateAI Careers

🔥 Small Yet Mighty: Microsoft's Phi-3 AI Model Rivals GPT-3.5

🚀 Microsoft has launched Phi-3-Mini, a compact AI model designed to run on smartphones and laptops, marking a significant advancement in mobile AI capabilities.

🏃‍♀️ This 3.8 billion parameter model, trained on 3.3 trillion tokens, is competitive with larger models like Mixtral 8x7B and GPT-3.5, achieving impressive benchmarks on MMLU and MT-bench. It offers potential for content creation and social media post generation.

🤷‍♀️ The model can be quantized to 4-bits, using ≈ 1.8GB of memory on phones, achieving 12 tokens/s natively on iPhone 14. This development is seen as a step towards more efficient and accessible AI capabilities on mobile devices, with the Phi-3 Technical Report describing a model that is competitive with Mixtral 8x7B in MMLU.

🙌 The model is also open-source, similar to Llama-3, and is expected to benefit from open-source contributions.

To deep dive into architecture of these models , and a detailed demo SUBSCRIBE 👇 to my YouTube channel AccelerateAICareers and watch my Popular LLM series!


If you like this content , ♻ share it with you friends!!

Here , in this shorts video, Sebastien Bubeck (VP, Generative AI Research, Microsoft AI) demoes Phi-3 mini: https://youtu.be/XoOvZS3wdic?si=EK0s_...

#ai #artificialintelligence #microsoft #microsoftAI #openai #chatgpt #chatgpt3 #chatgpt4 #largelanguagemodels #generative #generativeai #genai #deeplearning #machinelearning #datascience #ainews #ailearning #technology #tech #engineering #softwaredeveloper #software #python #technews #leaders #founders

1 year ago (edited) | [YT] | 4

AccelerateAI Careers

Today, Meta released Llama3. In context of that , I have created a Deep Dive video for Llama2. I am going to release a similar video on Llama3 very soon! If you have not done yet, SUBSCRIBE to my channel for similar content!

#llm #llms #AI #artificialintelligence #machinelearning #machinelearningbasics #generative #generativeai #deeplearning #foundationmodels #viral #generativemodels #llama2 #llama3 #largelanguagemodels #largelanguagemodel #popularllm #ailearning #aiconcepts #nlp #nlptutorial

1 year ago | [YT] | 4

AccelerateAI Careers

⚡️WhatsApp's ‪@meta‬ AI chat feature released!! ⚡️

👉It allows engaging conversations, question-asking, and searching of text , and synthesis of text and image responses!!

⚡️How to access and use it:
1. Open WhatsApp and navigate to the chat screen.
2. Click the "New Chat" button and select "Meta AI."
3. If prompted, read and accept the terms of service.
4. To chat with Meta AI, use @Meta AI ; to generate new information, Type "@" in the message field and select "/imagine".

💥This is such a customer oriented approach and a good use of Generative AI as one really does not need to go install chatGPT or PerplexityAI or search in web browsers when you can chat in your messaging application itself… This could change the game for search engines as well 🌟

What do you think about this newly added Chat with MetaAI feature in your whatsapp app? Do you think it is the right move? Answer in comments!! SUBSCRIBE to this channel for similar news in AI and Generative AI.

#whatsapp #meta #ai #ml #generative #ainews #artificialintelligence #searchengine #generativeainews #genai #technology #engineering #innovation ‪@AccelerateAICareers‬ ‪@meta‬ ‪@WhatsApp‬

1 year ago (edited) | [YT] | 4