← νŠΈλ Œλ”©

Tangerine IndexλŠ” ChatGPT 이후 λ‚˜μ˜¨ ν”„λ‘œλ•νŠΈμ™€ μ‚¬λžŒλ“€μ˜ λ°˜μ‘μ„ μ‹£μŠ΅λ‹ˆλ‹€.

minimind

github.com

🧠 Train a 64M-parameter LLM from scratch in just 2h!

μ˜€ν”ˆμ†ŒμŠ€

github.com λ°©λ¬Έμ™ΈλΆ€ μ‚¬μ΄νŠΈGitHub64,980 μŠ€νƒ€μ™ΈλΆ€ μ‚¬μ΄νŠΈ

λ°˜μ‘ (1)

λΉ„μŠ·ν•œ ν”„λ‘œλ•νŠΈ

μ„€λͺ…이 λΉ„μŠ·ν•œ 순

  1. minimind-v

    πŸ‘€ Train a 65M-parameter VLM from scratch in just 2h!

    github.comμ˜€ν”ˆμ†ŒμŠ€μ΅œκ·Ό λ°˜μ‘

    GitHub
    μŠ€νƒ€ 8,822

  2. llm-internals

    Learn LLM internals step by step - from tokenization to attention to inference optimization.

    github.comμ˜€ν”ˆμ†ŒμŠ€μ΅œκ·Ό λ°˜μ‘

    GitHub
    μŠ€νƒ€ 1,770

  3. gemma

    Gemma open-weight LLM library, from Google DeepMind

    github.comλΌμ΄λΈŒλŸ¬λ¦¬μ˜€ν”ˆμ†ŒμŠ€μ΅œκ·Ό λ°˜μ‘

    GitHub
    μŠ€νƒ€ 5,838

  4. Medusa

    Medusa: Simple Framework for Accelerating LLM Generation with Multiple Decoding Heads

    github.comλΌμ΄λΈŒλŸ¬λ¦¬μ˜€ν”ˆμ†ŒμŠ€μ΅œκ·Ό λ°˜μ‘

    GitHub
    μŠ€νƒ€ 2,774

  5. tokenspeed

    TokenSpeed is a speed-of-light LLM inference engine.

    github.comμ˜€ν”ˆμ†ŒμŠ€μ΅œκ·Ό λ°˜μ‘

    GitHub
    μŠ€νƒ€ 2,885