Hugging Face's logo

syedazeez
/
deepseek-v3-compact-proxy

Text Generation
PyTorch
English
deepseek
deepseek-v3
mixture-of-experts
multi-head-latent-attention
multi-token-prediction
research
Model card Files
xet
 Community
deepseek-v3-compact-proxy

Commit History

  1. Point the source link at the AI-Yatra organisation

    syedazeez
    Verified signature
    45f005c

  2. List the Gate V shakedown checkpoint in the index

    syedazeez
    Verified signature
    da8ff41

  3. Add Gate V WikiText-103 shakedown checkpoint (step 8100, 87.57 val PPL)

    syedazeez
    Verified signature
    a435c67

  1. Gate W: 2-epoch WikiText-103 checkpoint, 41.35 validation perplexity

    syedazeez
    Verified signature
    f110bce

  2. Add README.md for the Gate W WikiText-103 checkpoint

    syedazeez
    Verified signature
    585f729

  3. Add tokenizer_wikitext103.json for the Gate W WikiText-103 checkpoint

    syedazeez
    Verified signature
    f3567e3

  1. Add YAML metadata and Apache 2.0 license to model card

    syedazeez
    Verified signature
    c6f27bc

  2. Upload all gate checkpoints (Configuration A and B)

    syedazeez
    Verified signature
    475caef

  3. initial commit

    syedazeez
    Verified signature
    e58dfd3