All pull requests

Pull requests

• docs: add chat template and streaming example; fix training snippet

  #387·mr-shakib opened Oct 20, 2026·

  ·

• Load single-file checkpoints once upfront during splitting and handle non-numeric shard names

  #380·ALANDVO opened Oct 12, 2026·

  ·
  1

• Resume interrupted layer splitting without rewriting saved shards

  #378·mvanhorn opened Oct 11, 2026·2 tasks done·

  ·
  1

• Add dry-run disk and VRAM estimates before weight download

  #377·mvanhorn opened Oct 11, 2026·4 tasks done·

  ·
  1

• Enable ROCm inference with Strix Halo validation

  #373·sa-faizal opened Oct 6, 2026·

  ·
  1

  2

• Add CLI and interactive terminal chat

  #358·prathamc00 opened Sep 4, 2026·4 tasks done·

  ·
  7

• feat: Add fixed-slot GPU expert streaming for PhiMoE

  #350·DevElCuy opened Aug 24, 2026·

  ·
  1

• feat: support prequantized BitsAndBytes streaming with Transformers 5

  #347·DevElCuy opened Aug 23, 2026·

  ·
  1

• Support nested Gemma 4 checkpoint layout

  #346·ousamabenyounes opened Aug 22, 2026·

  ·
  1

  2

• fix: robust layer-count parsing for nested checkpoint prefixes (complements #336)

  #339·HEETMEHTA18 opened Aug 9, 2026·

  ·
  1

  12

• fix: handle missing files gracefully in remove_real_and_linked_file (#297)

  #337·bhumitschaudhry opened Aug 7, 2026·

  ·
  1

  6

• docs: clarify compression performance tradeoffs

  #334·aryansk opened Aug 6, 2026·

  ·
  1

  3

• docs: add Russian README translation

  #332·MonteNegroX opened Aug 5, 2026·

  ·

• docs: list Kimi K3 in Supported Models (#316)

  #327·evanclan opened Aug 4, 2026·3 of 4 tasks·

  ·
  1

• feat: AIRLLM_FULL_LOAD bypass — skip layer streaming when model fits in VRAM

  #324·he1074937561-a11y opened Aug 3, 2026·

  ·
  4

• feat: add lightweight CPU-only CI/test harness

  #322·ardiannurcahya opened Aug 2, 2026·

  ·
  1

• Fix the unit tests so they can actually run

  #315·Hotragn opened Jul 28, 2026·

  ·
  2

• Use the model's real generation config for custom architectures

  #313·Hotragn opened Jul 24, 2026·

  ·

• Fix 4-bit disk-space check demanding ~3.5x the model size

  #309·Hotragn opened Jul 23, 2026·

  ·

• Default to trust_remote_code=False, fall back only when required (addresses #293)

  #306·Hotragn opened Jul 20, 2026·

  ·
  1

• Prefer safetensors over pickle .bin when splitting model layers (fixes #296)

  #305·Hotragn opened Jul 20, 2026·

  ·
  1

• Fix remove_real_and_linked_file double-deleting normal files

  #303·kapil971390 opened Jul 19, 2026·

  ·
  1

  1

• Quality-of-life fork: refactored AutoModel, pyproject.toml, unit tests, CI

  #291·Doe-Johnevro opened Jun 4, 2026·

  ·

• Fix per-layer streaming for transformers >=4.46

  #288·einyx opened May 21, 2026·

  ·

• macOS MLX path: module reuse, mx.fast.* speedups, native 4bit

  #286·MartinLilt opened May 8, 2026·3 of 4 tasks·

  ·