Skip to content

Latest commit

 

History

History
114 lines (87 loc) · 23.8 KB

README.md

File metadata and controls

114 lines (87 loc) · 23.8 KB

llmducation

Staying up to date with Large Language Models (no hype).

This is my plan for staying excited about large language models:

  1. Stop worrying about missing the latest trendy python framework.

  2. Make a reading list which covers all of the important LLM innovations (both academia and industry) and work through the list. I want the list to contain both papers (academia) and commentary on (successful) practical applications of LLMs.

  3. Subscribe to email newsletters, LinkedIn, YouTube and Twitter accounts of significant LLM innovators and commentators (avoiding sources focusing on hype and speculation).

Here is my (constantly evolving) reading list:

Status Title Value Year Topic Type Notes Link(s)
✅ read ReAct: Synergizing Reasoning and Acting in Language Models ★★★★★ 2023 LLM Prompting paper Compelling and readable https://arxiv.org/abs/2210.03629
✅ read Self-consistency improves chain of thought reasoning in language models ★★★★☆ 2022 LLM Prompting paper https://arxiv.org/abs/2203.11171
✅ read Hermes: A Text-to-SQL solution at Swiggy ★★★★★ blog https://bytes.swiggy.com/hermes-a-text-to-sql-solution-at-swiggy-81573fb4fb6e
✅ read A survey on large language model based autonomous agents ★★★★★ 2024 paper https://link.springer.com/article/10.1007/s11704-024-40231-1
✅ read Reflexion: Language Agents with Verbal Reinforcement Learning ★★★★★ 2023 LLM Agents paper Simple but very powerful concept https://arxiv.org/abs/2303.11366
✅ read RAGChecker: A Fine-grained Framework for Diagnosing Retrieval-Augmented Generation ★★★★★ 2024 LLM Evaluation paper very convincing and elegant framework. Very thorough research https://arxiv.org/abs/2408.08067
✅ read RefChecker: Reference-based Fine-grained Hallucination Checker and Benchmark for Large Language Models 2024 LLM Evaluation paper this framework is also used in the RAGChecker paper https://arxiv.org/abs/2405.14486
✅ read Cognitive Architectures for Language Agents ★★★★★ 2023 LLM Agents paper notes https://arxiv.org/abs/2309.02427
✅ read Scaling Test Time Compute with Open Models ★★★★★ 2024 Test Time Compute blog Much more readable than the original paper https://huggingface.co/spaces/HuggingFaceH4/blogpost-scaling-test-time-compute
✅ read Unleashing the Emergent Cognitive Synergy in Large Language Models: A Task-Solving Agent through Multi-Persona Self-Collaboration ★★★★★ 2024 reasoning paper I loved this paper! Such a straightforward but creative concept https://arxiv.org/abs/2307.05300
❌ not started KAG: Boosting LLMs in Professional Domains via Knowledge Augmented Generation 2024 RAG paper
✅ read Reasoning over Uncertain Text by Generative Large Language Models ★★★☆☆ 2024 paper Less applied than I'd hoped (more theoretical) but still quite interesting https://arxiv.org/abs/2402.09614v3
❌ not started Understanding Multimodal LLMs 2024 Multi-Modal LLMs blog notes https://www.linkedin.com/pulse/understanding-multimodal-llms-sebastian-raschka-phd-t7h5c
❌ not started Large Language Models Understand and Can Be Enhanced by Emotional Stimuli 2023 LLM Prompting paper
❌ not started The Llama 3 Herd of Models 2024 Foundation models paper
◐ partially read Build a Large Language Model (from scratch) ★★★★★ LLM Architecture book If I could read only 1 thing, this would be it
✅ read Orchestrating Agents: Routines and Handoffs ★★★★☆ 2024 LLM Agents blog Very clear. Nice python code examples with no additional frameworks required https://cookbook.openai.com/examples/orchestrating_agents
✅ read Endless Jailbreaks with Bijection Learning ★★★★★ 2024 LLM Security paper very simple but powerful technique https://arxiv.org/abs/2410.01294
✅ read Multi-expert Prompting Improves Reliability, Safety, and Usefulness of Large Language Models ★★★★☆ 2024 paper Very novel, effective and interpretable technique https://arxiv.org/abs/2411.00492
◐ partially read Fine-tuning Embedding Models for RAG ★★★★★ 2024 Embeddings blog Impressive performance gain. Contains lots of useful code https://www.philschmid.de/fine-tune-embedding-model-for-rag
◐ partially read Graph Retrieval-Augmented Generation: A Survey ★★★☆☆ 2024 RAG paper notes https://arxiv.org/abs/2408.08921
❌ not started Great, Now Write an Article About That: The Crescendo Multi-Turn LLM Jailbreak Attack 2024 LLM Security paper https://arxiv.org/abs/2404.01833
✅ read BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding ★★★★☆ 2018 Foundation Models paper https://arxiv.org/abs/1810.04805
❌ not started Chain-of-Thought Reasoning Without Prompting year topic type notes https://arxiv.org/abs/2402.10200
✅ read o1 isn't a chat model (and that's the point) ★★★★☆ 2025 Test-time compute blog https://www.latent.space/p/o1-skill-issue
❌ not started Tree of Thoughts: Deliberate Problem Solving with Large Language Models ★★☆☆☆ year topic type notes https://arxiv.org/abs/2305.10601
❌ not started Graph of Thoughts: Solving Elaborate Problems with Large Language Models 2023 LLM Agents paper notes https://arxiv.org/abs/2308.09687
❌ not started GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models 2024 LLM Limitations paper https://arxiv.org/abs/2410.05229v1
❌ not started From Local to Global: A Graph RAG Approach to Query-Focused Summarization 2024 RAG paper notes links
◐ partially read What We've Learned From a Year of Building with LLMs 2024 llm-ops blog https://applied-llms.org
❌ not started Large Language Models are Zero-Shot Reasoners 2022 LLM Prompting paper notes https://arxiv.org/abs/2205.11916
❌ not started Planning with Large Language Models via Corrective Re-prompting 2022 topic paper notes https://openreview.net/pdf?id=cMDMRBe1TKs
❌ not started Evaluating and enhancing probabilistic reasoning in language models 2024 blog notes https://research.google/blog/evaluating-and-enhancing-probabilistic-reasoning-in-language-models/
❌ not started ReWOO: Decoupling Reasoning from Observations for Efficient Augmented Language Models 2023 LLM Agents paper notes https://arxiv.org/abs/2305.18323
❌ not started Sparse Priming Representations 2023 github repo https://github.com/daveshap/SparsePrimingRepresentations
https://www.youtube.com/watch?v=YjdmYCd6y0M
❌ not started BOLAA: Benchmarking and Orchestrating LLM-augmented Autonomous Agents 2023 LLM Agents paper https://arxiv.org/abs/2308.05960
https://github.com/salesforce/BOLAA
◐ partially read ML and LLM system design: 450 case studies to learn from ★★★★★ 2024 llm-ops website notes https://www.evidentlyai.com/ml-system-design
❌ not started Your RAG application is a communication system 2024 RAG blog notes https://superlinked.com/vectorhub/articles/rag-application-communication-system
❌ not started Reader-LM: Small Language Models for Cleaning and Converting HTML to Markdown 2024 blog notes https://jina.ai/news/reader-lm-small-language-models-for-cleaning-and-converting-html-to-markdown/
❌ not started Writing in the Margins: Better Inference Pattern for Long Context Retrieval 2024 RAG paper notes https://www.arxiv.org/abs/2408.14906
❌ not started Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters 2024 paper notes https://arxiv.org/abs/2408.03314
❌ not started LightRAG: Simple and Fast Retrieval-Augmented Generation 2024 RAG paper notes https://arxiv.org/abs/2410.05779
❌ not started GPT-4 Technical Report 2024 LLM Architecture paper notes https://arxiv.org/abs/2303.08774
❌ not started Sparks of Artificial General Intelligence: Early experiments with GPT-4 2023 AGI paper notes https://arxiv.org/abs/2303.12712
❌ not started A Survey of Large Language Models ongoing (started 2023) paper (survey) notes https://arxiv.org/abs/2303.18223
❌ not started QLoRA: Efficient Finetuning of Quantized LLMs 2023 Quantization paper notes https://arxiv.org/abs/2305.14314
❌ not started Mamba: Linear-Time Sequence Modeling with Selective State Spaces 2024 LLM Architecture paper notes https://arxiv.org/abs/2312.00752
❌ not started Towards Expert-Level Medical Question Answering with Large Language Models 2023 paper notes https://arxiv.org/abs/2305.09617
❌ not started ZEBRA: Zero-Shot Example-Based Retrieval Augmentation for Commonsense Question Answering 2024 RAG paper notes https://arxiv.org/abs/2410.05077
◐ partially read Two Tales of Persona in LLMs: A Survey of Role-Playing and Personalization 2024 LLM Persona paper (survey) https://arxiv.org/abs/2406.01171
◐ partially read Quantifying the Persona Effect in LLM Simulations 2024 LLM Persona paper notes https://arxiv.org/abs/2402.10811
❌ not started Looking Inward: Language Models Can Learn About Themselves by Introspection 2024 paper notes https://arxiv.org/abs/2410.13787
❌ not started Who Validates the Validators? Aligning LLM-Assisted Evaluation of LLM Outputs with Human Preferences 2024 LLM Evaluation paper notes https://arxiv.org/abs/2404.12272
✅❌templaterow title ★★☆☆☆ year topic type notes links

Also to check out:

LLM sampling techniques:

  • temperature sampling (Ackley et al., 1985; Ficler & Goldberg, 2017)

  • top-k sampling (Fan et al., 2018; Holtzman et al., 2018; Radford et al., 2019)

  • nucleus sampling (Holtzman et al., 2020)

Subscribed to/followed:

  • Data Elixir

  • Andrej Karpathy

  • Sebastian Raschka

  • AI Tidbits

  • AlphaSignal

  • Data Science Weekly

  • The Batch (Andrew NG)

Interesting projects:

Project Name Description Link(s)
optillm <https://github.com/codelion/optillm

References and Resources: