Commit History

  1. Increase server timeout to 600s for large model experiments

    AdriBat1
    044ff0a
  2. Add DeepSeek-Lite Protocol: 50M params, FineWeb-Edu, TikToken, BFloat16

    AdriBat1
    6ec2818
  3. Add Tower of Babel (V3) experiment: 120-layer, gradient monitoring, val loss

    AdriBat1
    938275a

  1. Add Deep-NanoGPT experiment (Phase 1 & 2): resumable training, inference, 72-layer models

    AdriBat1
    671ce97
  2. Add and verify LLM training and inference examples

    AdriBat1
    5f654f8
  3. Implement and verify remote persistence workflow

    AdriBat1
    979e977
  4. Reorganize client files into examples, tests, and output folders

    AdriBat1
    1b272e7