Commits

Commit history

1. Commits on Sep 30, 2026

   • add GLM-5.3-Flash (GLM5-Next) support (#27773)

     Show description for 649dcb1
     

     4 peopleauthored
     649dcb1View commit details

     Copy full SHA for 649dcb1

     View code at this pointBrowse repository at this point
2. Commits on Sep 1, 2026

   • quantize: row-slab stream to avoid thread starvation (#27830)

     ngxson

     authored
     1b89a43View commit details

     Copy full SHA for 1b89a43

     View code at this pointBrowse repository at this point
3. Commits on Aug 27, 2026

   • model: add Qwen3.8-Flash-Next (qwen4exp) (#27742)

     Show description for 6c84c7d
     

     5 peopleauthored
     6c84c7dView commit details

     Copy full SHA for 6c84c7d

     View code at this pointBrowse repository at this point

   • quantize: cap working memory size to avoid loading big tensors onto RAM (#27795)

     ngxson

     authored
     732707dView commit details

     Copy full SHA for 732707d

     View code at this pointBrowse repository at this point
4. Commits on Aug 18, 2026

   • quant : Optimise memory usage by evicting weights after processing each layer (#22877)

     Show description for 0596704
     EAddario

     authored
     0596704View commit details

     Copy full SHA for 0596704

     View code at this pointBrowse repository at this point
5. Commits on Aug 17, 2026

   • model : BailingMoE3 Support (#26608)

     Show description for 3733366
     

     aetherbird

     and

     Dyluhn

     authored
     3733366View commit details

     Copy full SHA for 3733366

     View code at this pointBrowse repository at this point
6. Commits on Jul 31, 2026

   • llama : load MTP tensors only if they are really used (#26296)

     Show description for 82dbc4f
     

     fairydreaming

     and

     sszymczy

     authored
     82dbc4fView commit details

     Copy full SHA for 82dbc4f

     View code at this pointBrowse repository at this point
7. Commits on Jul 27, 2026

   • mtmd: support MiMo-V2.5 audio input (RVQ-based model) (#26190)

     Show description for 1cbfd19
     ngxson

     authored
     1cbfd19View commit details

     Copy full SHA for 1cbfd19

     View code at this pointBrowse repository at this point
8. Commits on Jul 26, 2026

   • Keep Minimax's indexer tensors at F32 for speed and accuracy (#26144)

     Show description for 7657a6c
     bartowski1182

     authored
     7657a6cView commit details

     Copy full SHA for 7657a6c

     View code at this pointBrowse repository at this point
9. Commits on Jul 24, 2026

   • llama: various bug fixes (#26051)

     ngxson

     authored
     298219fView commit details

     Copy full SHA for 298219f

     View code at this pointBrowse repository at this point
10. Commits on Jul 23, 2026

    • args: refactor mlock/mmap/directio into load-mode (#20834)

      Show description for e6dd0e2
      taronaeo

      authored
      e6dd0e2View commit details

      Copy full SHA for e6dd0e2

      View code at this pointBrowse repository at this point
11. Commits on Jul 18, 2026

    • llama-quant : exclude i32 ffn_gate_tid2eid routing table from quantization (#25787)

      Show description for 4937ca8
      devYRPauli

      authored
      4937ca8View commit details

      Copy full SHA for 4937ca8

      View code at this pointBrowse repository at this point
12. Commits on Jul 16, 2026

    • quant : allow using manual tensor types with --pure (#25716)

      ggerganov

      authored
      56d6e9dView commit details

      Copy full SHA for 56d6e9d

      View code at this pointBrowse repository at this point
13. Commits on Jul 7, 2026

    • Add Q2_0 quantization: type definition and CPU backend (#24448)

      khosravipasha

      authored
      bec4772View commit details

      Copy full SHA for bec4772

      View code at this pointBrowse repository at this point
14. Commits on Jun 25, 2026

    • quant : fix quantizing moe with mtp (#24986)

      CISC

      authored
      b3ce5ceView commit details

      Copy full SHA for b3ce5ce

      View code at this pointBrowse repository at this point
15. Commits on Jun 20, 2026

    • llama : use LLM_KV for quantization_version & file_type (#24802)

      Show description for 84de01a
      angt

      authored
      84de01aView commit details

      Copy full SHA for 84de01a

      View code at this pointBrowse repository at this point
16. Commits on Jun 5, 2026

    • hparams : refactor `hparams.n_layer` (#24060)

      Show description for 7acb4e8
      

      ggerganov

      and

      CISC

      authored
      7acb4e8View commit details

      Copy full SHA for 7acb4e8

      View code at this pointBrowse repository at this point
17. Commits on May 4, 2026

    • model: move `load_hparams` and `load_tensors` to per-model definition (#22004)

      Show description for 994118a
      ngxson

      authored
      994118aView commit details

      Copy full SHA for 994118a

      View code at this pointBrowse repository at this point
18. Commits on May 1, 2026

    • llama-quant : fix `--tensor-type` when default `qtype` is overriden (#22572)

      Show description for b97ebdc
      ddh0

      authored
      b97ebdcView commit details

      Copy full SHA for b97ebdc

      View code at this pointBrowse repository at this point
19. Commits on Apr 25, 2026

    • llama-quant : default ftype param `Q5_1` --> `Q8_0` (#20828)

      Show description for 9d34231
      ddh0

      authored
      9d34231View commit details

      Copy full SHA for 9d34231

      View code at this pointBrowse repository at this point
20. Commits on Apr 6, 2026

    • ggml: add Q1_0 1-bit quantization support (CPU) (#21273)

      Show description for 2e1f0a8
      

      khosravipasha

      and

      CISC

      authored
      2e1f0a8View commit details

      Copy full SHA for 2e1f0a8

      View code at this pointBrowse repository at this point
21. Commits on Apr 2, 2026

    • tests : add unit test coverage for llama_tensor_get_type (#20112)

      Show description for 7992aa7
      

      3 peopleauthored
      7992aa7View commit details

      Copy full SHA for 7992aa7

      View code at this pointBrowse repository at this point
22. Commits on Apr 1, 2026

    • llama : refactor llama_model_quantize_params to expose a pure C interface (#20346)

      Show description for 4951250
      

      EAddario

      and

      ggerganov

      authored
      4951250View commit details

      Copy full SHA for 4951250

      View code at this pointBrowse repository at this point
23. Commits on Mar 26, 2026

    • mtmd: fix "v.patch_embd" quant and unsupported im2col ops on Metal for deepseek-ocr (#21027)

      Show description for 1743d98
      

      sfallah

      and

      CISC

      authored
      1743d98View commit details

      Copy full SHA for 1743d98

      View code at this pointBrowse repository at this point
24. Commits on Mar 25, 2026

    • mtmd: Add DeepSeekOCR Support (#17400)

      Show description for a970515
      

      4 peopleauthored
      a970515View commit details

      Copy full SHA for a970515

      View code at this pointBrowse repository at this point

    • llama: fix llama-model-saver (#20503)

      Show description for 36dafba
      

      JohannesGaessler

      and

      Siddhesh2377

      authored
      36dafbaView commit details

      Copy full SHA for 36dafba

      View code at this pointBrowse repository at this point
25. Commits on Mar 10, 2026

    • llama-quant : correct `n_attention_wv` usage (#20357)

      Show description for 10e5b14
      ddh0

      authored
      10e5b14View commit details

      Copy full SHA for 10e5b14

      View code at this pointBrowse repository at this point

    • llama-quant : fail early on missing imatrix, refactor type selection, code cleanup (#19770)

      Show description for 1dab5f5
      ddh0

      authored
      1dab5f5View commit details

      Copy full SHA for 1dab5f5

      View code at this pointBrowse repository at this point
26. Commits on Mar 9, 2026

    • llama-quant : left-align tensor names in output (#20117)

      ddh0

      authored
      b518195View commit details

      Copy full SHA for b518195

      View code at this pointBrowse repository at this point
27. Commits on Mar 8, 2026

    • llama: end-to-end tests (#19802)

      Show description for a976ff0
      JohannesGaessler

      authored
      a976ff0View commit details

      Copy full SHA for a976ff0

      View code at this pointBrowse repository at this point
28. Commits on Feb 20, 2026

    • quantize : add --dry-run option (#19526)

      Show description for 492bc31
      

      ddh0

      and

      CISC

      authored
      492bc31View commit details

      Copy full SHA for 492bc31

      View code at this pointBrowse repository at this point
29. Commits on Feb 6, 2026

    • Kimi-Linear support (backend agnostic + MLA KV cache) (#18755)

      Show description for 3688c4f
      

      ymcki

      and

      pwilkin

      authored
      3688c4fView commit details

      Copy full SHA for 3688c4f

      View code at this pointBrowse repository at this point
30. Commits on Jan 28, 2026

    • llama : disable Direct IO by default (#19109)

      Show description for c5c64f7
      ggerganov

      authored
      c5c64f7View commit details

      Copy full SHA for c5c64f7

      View code at this pointBrowse repository at this point
31. Commits on Jan 22, 2026

    • quant : manual overrides of tensor types take precedence (#18952)

      ggerganov

      authored
      0e4ebebView commit details

      Copy full SHA for 0e4ebeb

      View code at this pointBrowse repository at this point
32. Commits on Jan 8, 2026

    • llama : add `use_direct_io` flag for model loading (#18166)

      Show description for 2038101
      

      3 peopleauthored
      2038101View commit details

      Copy full SHA for 2038101

      View code at this pointBrowse repository at this point