harleyszhang/llm_note

• highabout#1

  Clarify the repository's "About" description

  Why:

  CURRENT

  LLM notes, including model inference, transformer model structure, and llm framework code analysis notes.

  COPY-PASTE FIX

  Copy

  A comprehensive course and notes on building and optimizing LLM inference frameworks, covering model inference, transformer architecture, HPC programming (Triton/CUDA), and framework code analysis (vLLM).

  OpenDoneDismiss
• highlicense#2

  Add a LICENSE file to the repository

  Why:

  CURRENT

  (no LICENSE file detected — the repo has no recognizable license)

  COPY-PASTE FIX

  Copy

  Choose and add a standard SPDX-compliant LICENSE file (e.g., MIT, Apache-2.0, GPL-3.0) to the repository root.

  OpenDoneDismiss
• mediumreadme#3

  Strengthen the README's opening statement to emphasize the course and framework building aspects

  Why:

  CURRENT

  LLM notes, including model inference, hpc programming note, transformer model structure, and vllm framework code analysis notes.

  COPY-PASTE FIX

  Copy

  This repository provides comprehensive notes and a self-made course on building and optimizing LLM inference frameworks, covering model inference, transformer architecture, HPC programming (Triton/CUDA), and in-depth code analysis of frameworks like vLLM.

  OpenDoneDismiss

Category GEO backends resolved for this scan: google/gemini-2.5-flash, deepseek/deepseek-v4-flash

• CATEGORY QUERY

  How can I build a performant LLM inference framework using Triton and PyTorch?

  you: not recommended
  AI recommended (in order):

  1. AWQ
  2. GPTQ
  3. bitsandbytes
  4. FlashAttention-2
  5. xFormers
  6. Triton
  7. PyTorch
  8. `torch.compile` (Dynamo)
  9. `torch.fx`
  10. vLLM
  11. TensorRT-LLM
  12. DeepSpeed-MII (Model Inference Interface)

  AI recommended 12 alternatives but never named harleyszhang/llm_note. This is the gap to close.

  Show full AI answer
• CATEGORY QUERY

  Seeking resources to understand and optimize LLM inference performance, especially KV cache and quantization.

  you: not recommended
  AI recommended (in order):

  1. transformers (huggingface/transformers)
  2. optimum (huggingface/optimum)
  3. ONNX Runtime (microsoft/onnxruntime)
  4. OpenVINO (openvinotoolkit/openvino)
  5. NVIDIA TensorRT-LLM (NVIDIA/TensorRT-LLM)
  6. VLLM (vllm-project/vllm)
  7. GPTQ (IST-DASLab/gptq)
  8. AWQ (mit-han-lab/awq)
  9. BitsAndBytes (TimDettmers/bitsandbytes)
  10. DeepSpeed-MII (microsoft/DeepSpeed)

  AI recommended 10 alternatives but never named harleyszhang/llm_note. This is the gap to close.

  Show full AI answer

• Metadata completeness

  warn

  Suggestion:

• README presence

  pass

• Compared to common alternatives in this category, what is the core differentiator of harleyszhang/llm_note?

  pass
  ✓AI named harleyszhang/llm_note explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

• If a team adopts harleyszhang/llm_note in production, what risks or prerequisites should they evaluate first?

  pass
  ✓AI named harleyszhang/llm_note explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?

• In one sentence, what problem does the repo harleyszhang/llm_note solve, and who is the primary audience?

  pass
  ✓AI named harleyszhang/llm_note explicitly

  AI answers can be confidently wrong. Read for accuracy: does it match your actual tech stack, audience, and differentiator?