← All tracks

Self-paced course

Paper Breakdowns

The “What the Paper Actually Says” series — close readings of the systems and ML papers that matter, focused on the mechanism, the numbers, and when not to use the idea.

41 lessons~8h totalFree
Start learning →
0 / 41 lessons complete

Curriculum

  1. Cassandra: What the Paper Actually Says21 min read
  2. EAGLE: Speculative Decoding with Feature-Level Prediction — What the Paper Actually Says12 min read
  3. The Llama 3 Herd of Models: What the Paper Actually Says11 min read
  4. LLM.int8(): What the 8-bit Matrix Multiplication Paper Actually Says10 min read
  5. Mooncake: What the KV-Cache-Centric Disaggregated Serving Paper Actually Says14 min read
  6. Pregel: What the Large-Scale Graph Processing Paper Actually Says14 min read
  7. Switch Transformers: What the Sparse MoE Scaling Paper Actually Says13 min read
  8. Titans: What the Test-Time Memorization Paper Actually Says9 min read
  9. T5: What the Text-to-Text Paper Actually Says13 min read
  10. SGLang and RadixAttention: What the Paper Actually Says11 min read
  11. SARATHI: What the Chunked-Prefill Paper Actually Says11 min read
  12. Mixture of Depths: What the Paper Actually Says14 min read
  13. MapReduce: What the Google Paper Actually Says12 min read
  14. Mamba: What the Selective State Space Paper Actually Says11 min read
  15. Kafka: What the Original Paper Actually Says16 min read
  16. H2O: Heavy-Hitter Oracle for KV Cache Eviction — What the Paper Actually Says14 min read
  17. DeepSeek-V3: What the Frontier-on-a-Budget Paper Actually Says13 min read
  18. Dapper: What Google's Distributed Tracing Paper Actually Says13 min read
  19. Toolformer: What the Paper Actually Says11 min read
  20. MegaScale: What ByteDance's 12,288-GPU Training Paper Actually Says14 min read
  21. LoRA: What the Low-Rank Adaptation Paper Actually Says10 min read
  22. HNSW: What the Vector Similarity Search Paper Actually Says13 min read
  23. GFS: What the Google File System Paper Actually Says16 min read
  24. Generative Agents: What the Paper Actually Says13 min read
  25. FrugalGPT: What the Paper Actually Says15 min read
  26. DistServe: Disaggregating Prefill and Decoding for Goodput-Optimized LLM Serving13 min read
  27. Bigtable: What the Distributed Storage Paper Actually Says12 min read
  28. Zanzibar: What Google's Authorization Paper Actually Says12 min read
  29. StreamingLLM: What the Attention Sink Paper Actually Says8 min read
  30. Spanner: What Google's Globally-Distributed Database Paper Actually Says11 min read
  31. Raft: What the Understandable Consensus Algorithm Paper Actually Says11 min read
  32. Multi-Token Prediction: What the Meta FAIR Paper Actually Says9 min read
  33. MapReduce: What the Paper Actually Says13 min read
  34. Llama 2: What the Open-Source RLHF Paper Actually Says11 min read
  35. BitNet b1.58: What the 1-bit LLM Paper Actually Says10 min read
  36. Tree of Thoughts: What the Paper Actually Says About LLM Search10 min read
  37. Self-RAG: What Adaptive Retrieval Actually Means in Production10 min read
  38. ReAct: What the Reasoning + Acting Paper Actually Says10 min read
  39. QLoRA: What the Paper Actually Says About Fine-Tuning on Consumer Hardware13 min read
  40. MLA: What the Multi-Head Latent Attention Paper Actually Says11 min read
  41. Mistral 7B: What the Sliding Window Attention Paper Actually Says9 min read