diff --git a/.nojekyll b/.nojekyll index ce7cd10..92e7611 100644 --- a/.nojekyll +++ b/.nojekyll @@ -1 +1 @@ -a4c0ee43 \ No newline at end of file +29e08d80 \ No newline at end of file diff --git a/lectures/w13-l01.html b/lectures/w13-l01.html index 963b6a3..b2d6300 100644 --- a/lectures/w13-l01.html +++ b/lectures/w13-l01.html @@ -416,28 +416,80 @@

In-Class

  • Datasets and Tasks (to train LLMs)
    • +
  • Model Training +
    • +
  • Prompt Optimization +
    • +
  • Constrained Language Generation +
    • +
  • Evaluation +
    • +
  • Applications and Design Patterns +
    • +
  • LLMs can not reason & plan +
    • + + +
    +

    Post-class

    +
      +
    1. Datasets and Tasks (to train LLMs) +
    2. Model Training
      3. @@ -463,12 +515,10 @@

      In-Class

    -
    -
    -

    Post-class

    -

    LLMs and Influence Functions

      @@ -510,9 +550,10 @@

      LLMs and Infl

      Full Courses

        -
      1. CIS7000 LLM Course @ UPenn by Prof. Mayur Naik. Covers many advanced topics
        2. +
      2. CIS7000 LLM Course @ UPenn by Prof. Mayur Naik. Covers many advanced topics.
      3. AIL821 LLMs Course @ IIT-D
      4. Deep Learning For NLP @ UW LING 574, Deep Learning For NLP, Prof. Shane @ UW, Spring’24.
        5. +
      5. Walk through the book Building LLMs from Scratch
      diff --git a/lectures/w13-l02.html b/lectures/w13-l02.html index b3b8f78..9e18866 100644 --- a/lectures/w13-l02.html +++ b/lectures/w13-l02.html @@ -429,6 +429,7 @@

      In-Class

    @@ -438,7 +439,7 @@

    Post-class

  • Training with Ray Train
  • Serving with Ray Serve, vLLM, ollama for serving
  • RAGs with Llamaindex, cognita, Langchain
    • -
  • Agens with LangGraph, AutoGen, CrewAI, LangGraph IDE blog
    • +
  • Agents with LangGraph, AutoGen, CrewAI, LangGraph IDE blog
    • diff --git a/lectures/w13-l03.html b/lectures/w13-l03.html index b73960b..afd0913 100644 --- a/lectures/w13-l03.html +++ b/lectures/w13-l03.html @@ -359,8 +359,7 @@

    Table of contents

  • Materials:
    • @@ -395,15 +394,44 @@

    Pre-work:

    1. LLM Intro
    2. LLM Ops
      3. +
    3. XAI Tutorial by Hima Lakkaraju, Julius Adebayo, Sameer Singh
      4. +
    4. UQ Tutorial by Balaji Lakshminarayanan
    -
    -

    In-Class

    -

    tbd

    -
    -
    -

    Post-class

    -

    tbd

    +
    +

    ML Engineering

    + +

    XAI

    + +

    UQ

    + +

    Security

    +
    diff --git a/search.json b/search.json index 85a559d..8785fe2 100644 --- a/search.json +++ b/search.json @@ -721,7 +721,7 @@ "href": "lectures/w13-l01.html#materials", "title": "13A: LLMs Introduction", "section": "", - "text": "Pre-work:\n\nLING571@UW Deep Learning For NLP, Prof. Shane at UW, Spring’24. Introduction, Word Vectors, Language Modeling\nCIS7000@UPenn LLMs, by Prof. Mayur Naik at UPenn, Fall’24, Background, Language Modeling\nAIL821@IIT-Delhi LLMs: Introduction and Recent Advances ELL881/AIL821, LLMs: Introduction and Advances @ IIT-Delhi, Fall’24.\nTransformers\n\nLLMs @ UPenn Part-1, Part-2\nLLMs: Introduction and Recent Advances @ IIT Delhi Module-5 on RNNs, Module-6 on Attention and Transformers\n\n\n\n\nIn-Class\nWe will follow the “Follow the data” approach to organize the content.\n\nQuick review of NLP and Deep Learning for NLP, pre- and post-GPT world.\n\nLecture 2 from AIL821 Introduction to NLP,\nLecture 3.1 from AIL821 Introduction to Language Models\n\nLLM Flow: (Quality) Datasets, Model Training (Pre-training, Alignment, Fine-tuning), Prompt Optimization, Constrained Language Generation, Evaluation.\nDatasets and Tasks (to train LLMs)\n\nLecture 7 from AIL821\nLIMA: less is more for alignment\nInstruction Tuning for Large Language Models: A Survey\nOLMo @ Allen AI\n\nModel Training\n\nPre-training\n\nLecture 12.1 from AIL821\nImproving Language Understanding by Generative Pre-Training\n\nAlignment\n\nLecture 12.2 from AIL821\nDirect Preference Optimization: Your Language Model is Secretly a Reward Model\n\nFine-tuning\n\nPerformance Efficient Fine-Tuning collection\nLecture: PEFT\nLecture: : Quantization and Pruning\nLoRA: Low-Rank Adaptation of Large Language Models\nQLoRA: Efficient Finetuning of Quantized LLMs\n\n\nPrompt Optimization\n\nThe Prompt Report\nChain-of-Thought\nTree-of-Thought\nSelf-Reflection\nSelf-Contrast\nThink before you Speak\n\nConstrained Language Generation\n\ncollection\nGuiding LLMs The Right Way: Fast, Non-Invasive Constrained Generation\n\nEvaluation\n\nScaling Evaluation of LLMs Yann Bubois, CIS 7000 LLM Course\n\nApplications and Design Patterns\n\nTools\n\nGorilla\n[Lecture 18.2 from AIL821] LLMs and Tools: Function Calling\n\nAgents\n\nLilian Wang’s blog on LLM Powered Autonomous Agents\n[Lecture 18.3 from AIL821] LLMs and Tools: Agentic\nAutoGen repo\nCrewAI repo\nLLM Agent papers collection \nSurvey: The Rise and Potential of Large Language Model Based Agents: A Survey\n\nRAG\n\nPaper from NVidia FACTS About Building Retrieval Augmented Generation-based Chatbots\nRetrieval-Augmented Generation for Large Language Models: A Survey Mar’24\nSearching for Best Practices in Retrieval-Augmented Generation Jul’24\n\n\nLLMs can not reason & plan\n\nLecture 19 from AIL821 Reasoning in LLMs\nLLMs Can’t Plan, But Can Help Planning in LLM-Modulo Frameworks\nGSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models\nLearning to reason with LLMs\nSystems-2 Collection Repo\n\n\n\n\nPost-class\n\nOther modules in AIL821 LLMs Course @ IIT-D\nAdvanced topics in CIS 7000 LLM Course\nWalk through the book Building LLMs from Scratch\n\n\nLLMs and Influence Functions\n\nStudying Large Language Model Generalization with Influence Functions\nDo Influence Functions Work on Large Language Models?\nTextGrad Automatic ‘’Differentiation’’ via Text, paper\n\n\n\nFull Courses\n\nCIS7000 LLM Course @ UPenn by Prof. Mayur Naik. Covers many advanced topics\nAIL821 LLMs Course @ IIT-D\nDeep Learning For NLP @ UW LING 574, Deep Learning For NLP, Prof. Shane @ UW, Spring’24.", + "text": "Pre-work:\n\nLING571@UW Deep Learning For NLP, Prof. Shane at UW, Spring’24. Introduction, Word Vectors, Language Modeling\nCIS7000@UPenn LLMs, by Prof. Mayur Naik at UPenn, Fall’24, Background, Language Modeling\nAIL821@IIT-Delhi LLMs: Introduction and Recent Advances ELL881/AIL821, LLMs: Introduction and Advances @ IIT-Delhi, Fall’24.\nTransformers\n\nLLMs @ UPenn Part-1, Part-2\nLLMs: Introduction and Recent Advances @ IIT Delhi Module-5 on RNNs, Module-6 on Attention and Transformers\n\n\n\n\nIn-Class\nWe will follow the “Follow the data” approach to organize the content.\n\nQuick review of NLP and Deep Learning for NLP, pre- and post-GPT world.\n\nLecture 2 from AIL821 Introduction to NLP,\nLecture 3.1 from AIL821 Introduction to Language Models\n\nLLM Flow: (Quality) Datasets, Model Training (Pre-training, Alignment, Fine-tuning), Prompt Optimization, Constrained Language Generation, Evaluation.\nDatasets and Tasks (to train LLMs)\n\nLecture 7 from AIL821\n\nModel Training\n\nPre-training\n\nLecture 12.1 from AIL821\n\nAlignment\n\nDirect Preference Optimization: Your Language Model is Secretly a Reward Model\n\nFine-tuning\n\nLoRA: Low-Rank Adaptation of Large Language Models\n\n\nPrompt Optimization\n\nChain-of-Thought\n\nConstrained Language Generation\n\nCollection\n\nEvaluation\n\nScaling Evaluation of LLMs Yann Bubois, CIS 7000 LLM Course\n\nApplications and Design Patterns\n\nTools\n\nGorilla\n\nAgents\n\nLilian Wang’s blog on LLM Powered Autonomous Agents\nAman’s blog on Agents\n\nRAG\n\nPaper from NVidia FACTS About Building Retrieval Augmented Generation-based Chatbots\n\n\nLLMs can not reason & plan\n\nLLMs Can’t Plan, But Can Help Planning in LLM-Modulo Frameworks\n\n\n\n\nPost-class\n\nDatasets and Tasks (to train LLMs)\n\nLIMA: less is more for alignment\nInstruction Tuning for Large Language Models: A Survey\nOLMo @ Allen AI - if you are interesting in all aspects of open-source LLM development.\n\nModel Training\n\nPre-training\n\nImproving Language Understanding by Generative Pre-Training\n\nAlignment\n\nLecture 12.2 from AIL821\n\nFine-tuning\n\nPerformance Efficient Fine-Tuning collection\nLecture: PEFT\nLecture: : Quantization and Pruning\nQLoRA: Efficient Finetuning of Quantized LLMs\n\n\nPrompt Optimization\n\nThe Prompt Report\nChain-of-Thought\nTree-of-Thought\nSelf-Reflection\nSelf-Contrast\nThink before you Speak\n\nConstrained Language Generation\n\ncollection\nGuiding LLMs The Right Way: Fast, Non-Invasive Constrained Generation\n\nEvaluation\n\nScaling Evaluation of LLMs Yann Bubois, CIS 7000 LLM Course\n\nApplications and Design Patterns\n\nTools\n\n[Lecture 18.2 from AIL821] LLMs and Tools: Function Calling\n\nAgents\n\n[Lecture 18.3 from AIL821] LLMs and Tools: Agentic\nAutoGen repo\nCrewAI repo\nLLM Agent papers collection \nSurvey: The Rise and Potential of Large Language Model Based Agents: A Survey\n\nRAG\n\nRetrieval-Augmented Generation for Large Language Models: A Survey Mar’24\nSearching for Best Practices in Retrieval-Augmented Generation Jul’24\n\n\nLLMs can not reason & plan\n\nLecture 19 from AIL821 Reasoning in LLMs\nGSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models\nLearning to reason with LLMs\nSystems-2 Collection Repo\n\n\n\nLLMs and Influence Functions\n\nStudying Large Language Model Generalization with Influence Functions\nDo Influence Functions Work on Large Language Models?\nTextGrad Automatic ‘’Differentiation’’ via Text, paper\n\n\n\nFull Courses\n\nCIS7000 LLM Course @ UPenn by Prof. Mayur Naik. Covers many advanced topics.\nAIL821 LLMs Course @ IIT-D\nDeep Learning For NLP @ UW LING 574, Deep Learning For NLP, Prof. Shane @ UW, Spring’24.\nWalk through the book Building LLMs from Scratch", "crumbs": [ "LLMs-Ops", "13A: LLMs Introduction" @@ -743,7 +743,7 @@ "href": "lectures/w13-l02.html#materials", "title": "13B: LLMs Ops", "section": "", - "text": "Pre-work:\n\nLLM Intro\n\n\n\nIn-Class\n\nBuild\n\nBuilding LLMs from Scratch repo\npre-training, fine-tuning, instruction fine-tuning\n\nPrompt Optimization\n\nDSPy, Intro Notebook\n\nStructured outputs in LLMs\n\nwith Instructor\nwith Outlines\nBlog comparing many libraries from structured output generation.\n\nEvaluation\n\nwith DeepEval\nwith ragas\nwith LangSmith\n\nApplications: RAGS\n\nBuilding RAG-based LLM Applications in Production blog, notebook by Gokul Mohandas and Philip Moritz\nAgentic RAG with LangGraph tutorial\n\nApplications: Agents\n\nwith LangGraph tutorials\nblog explaining LangGraph\n\n\n\n\nPost-class\n\nTraining with Ray Train\nServing with Ray Serve, vLLM, ollama for serving\nRAGs with Llamaindex, cognita, Langchain\nAgens with LangGraph, AutoGen, CrewAI, LangGraph IDE blog", + "text": "Pre-work:\n\nLLM Intro\n\n\n\nIn-Class\n\nBuild\n\nBuilding LLMs from Scratch repo\npre-training, fine-tuning, instruction fine-tuning\n\nPrompt Optimization\n\nDSPy, Intro Notebook\n\nStructured outputs in LLMs\n\nwith Instructor\nwith Outlines\nBlog comparing many libraries from structured output generation.\n\nEvaluation\n\nwith DeepEval\nwith ragas\nwith LangSmith\n\nApplications: RAGS\n\nBuilding RAG-based LLM Applications in Production blog, notebook by Gokul Mohandas and Philip Moritz\nAgentic RAG with LangGraph tutorial\n\nApplications: Agents\n\nwith LangGraph tutorials\nblog explaining LangGraph\nMagentic-One\n\n\n\n\nPost-class\n\nTraining with Ray Train\nServing with Ray Serve, vLLM, ollama for serving\nRAGs with Llamaindex, cognita, Langchain\nAgents with LangGraph, AutoGen, CrewAI, LangGraph IDE blog", "crumbs": [ "LLMs-Ops", "13B: LLMs Ops" @@ -765,7 +765,7 @@ "href": "lectures/w13-l03.html#materials", "title": "13C: Fullstack LLMs", "section": "", - "text": "Pre-work:\n\nLLM Intro\nLLM Ops\n\n\n\nIn-Class\ntbd\n\n\nPost-class\ntbd", + "text": "Pre-work:\n\nLLM Intro\nLLM Ops\nXAI Tutorial by Hima Lakkaraju, Julius Adebayo, Sameer Singh\nUQ Tutorial by Balaji Lakshminarayanan\n\n\n\nML Engineering\n\nLLaMA Stack - a full stack LLaMA-centered APIs for inference, safety, agentic system, among others.\nMLFlow LLMs - tool calling, agents, evaluation, RAGs, serving and more\nRay LLMs\nMLFlow Tracing observability for LLMs\nOthers popular stacks LlamaIndex, LangChain\nDeepEval\n\nXAI\n\nXAI @ Harvard, Spring’23, Explainable AI by Prof.Hima Lakkaraju\nPublications by Hima Lakkaraju\nLLMs for XAI\n\nCan Large Language Models Simplify Explainable AI\n\nXAI for LLMs\n\nStudying Large Language Model Generalization with Influence Functions\nDo Influence Functions Work on Large Language Models?\n\n\nUQ\n\nQuantifying Uncertainty in Natural Language Explanations of Large Language Models\nConformal Prediction with Large Language Models for Multi-Choice Question Answering code\n\nSecurity\n\nNeMO Gaurdrails\nLlaMA Gaurd 7B Model, paper", "crumbs": [ "LLMs-Ops", "13C: Fullstack LLMs" diff --git a/sitemap.xml b/sitemap.xml index ee84bb0..fe4d8c8 100644 --- a/sitemap.xml +++ b/sitemap.xml @@ -94,15 +94,15 @@ https://mlsquare.github.io/ai-839/lectures/w13-l01.html - 2024-11-07T04:20:24.960Z + 2024-11-11T11:59:08.120Z https://mlsquare.github.io/ai-839/lectures/w13-l02.html - 2024-11-07T05:05:23.522Z + 2024-11-11T13:01:30.939Z https://mlsquare.github.io/ai-839/lectures/w13-l03.html - 2024-11-07T04:00:07.470Z + 2024-11-11T13:10:48.897Z https://mlsquare.github.io/ai-839/homeworks.html