Skip to content
Permalink

Comparing changes

Choose two branches to see what’s changed or to start a new pull request. If you need to, you can also or learn more about diff comparisons.

Open a pull request

Create a new pull request by comparing changes across two branches. If you need to, you can also . Learn more about diff comparisons here.
base repository: ggml-org/llama.cpp
Failed to load repositories. Confirm that selected base ref is valid, then try again.
Loading
base: master
Choose a base ref
...
head repository: crow-cli/llama.cpp
Failed to load repositories. Confirm that selected head ref is valid, then try again.
Loading
compare: master
Choose a head ref
Checking mergeability… Don’t worry, you can still create the pull request.
  • 18 commits
  • 33 files changed
  • 2 contributors

Commits on Jun 21, 2026

  1. llama.cpp-mv: align with upstream b9743 (2026-06-22) 

    Weekly alignment: b9637 -> b9743. Output verified bit-identical.
Upstream changes: MoE w_s scaling, server refactoring, new models.

Assisted-by: Claude Sonnet
    @iz0eyj
    iz0eyj committed Jun 21, 2026
    Configuration menu
    Copy the full SHA
    d23d346 View commit details
    Browse the repository at this point in the history

Commits on Jun 28, 2026

  1. llama.cpp-mv: align with upstream b9833

    @iz0eyj
    iz0eyj committed Jun 28, 2026
    Configuration menu
    Copy the full SHA
    c9fa9ba View commit details
    Browse the repository at this point in the history

  2. llama.cpp-mv: update upstream tracker after b9833 alignment 

    Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jun 28, 2026
    Configuration menu
    Copy the full SHA
    369caad View commit details
    Browse the repository at this point in the history

Commits on Jun 30, 2026

  1. server: add --log-requests FNAME to log HTTP request/response JSON lines 

    Useful for debugging API clients and verifying embedding outputs.
Each line contains method, path, remote_addr, status, request body
and response body as JSON.

Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jun 30, 2026
    Configuration menu
    Copy the full SHA
    9aed05d View commit details
    Browse the repository at this point in the history

  2. docs: add --log-requests debugging note 

    Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jun 30, 2026
    Configuration menu
    Copy the full SHA
    03223f1 View commit details
    Browse the repository at this point in the history

Commits on Jul 5, 2026

  1. llama.cpp-mv: align with upstream b9876

    @iz0eyj
    iz0eyj committed Jul 5, 2026
    Configuration menu
    Copy the full SHA
    aaf96ef View commit details
    Browse the repository at this point in the history

  2. llama.cpp-mv: update upstream tracker after b9876 alignment 

    Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jul 5, 2026
    Configuration menu
    Copy the full SHA
    331906f View commit details
    Browse the repository at this point in the history

  3. llama.cpp-mv: -rea off disables reasoning extraction entirely 

    When -rea off is passed, also set reasoning_format to NONE so no
reasoning content is extracted or returned by the server.

Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jul 5, 2026
    Configuration menu
    Copy the full SHA
    9a58ffe View commit details
    Browse the repository at this point in the history

Commits on Jul 12, 2026

  1. llama.cpp-mv: align with upstream b9977

    @iz0eyj
    iz0eyj committed Jul 12, 2026
    Configuration menu
    Copy the full SHA
    10d40b0 View commit details
    Browse the repository at this point in the history

  2. llama.cpp-mv: update upstream tracker after b9977 alignment 

    Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jul 12, 2026
    Configuration menu
    Copy the full SHA
    727f3a5 View commit details
    Browse the repository at this point in the history

Commits on Jul 19, 2026

  1. llama.cpp-mv: add --fail-on-no-slot server option 

    When enabled, llama-server immediately returns HTTP 503 for new
completion/embedding/rerank requests if no slots are available,
instead of deferring them and risking OOM with concurrent traffic.

Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jul 19, 2026
    Configuration menu
    Copy the full SHA
    16bbce3 View commit details
    Browse the repository at this point in the history

  2. llama.cpp-mv: align with upstream b10068

    @iz0eyj
    iz0eyj committed Jul 19, 2026
    Configuration menu
    Copy the full SHA
    530297e View commit details
    Browse the repository at this point in the history

  3. docs: update bge-m3-sparse.md for b10068 alignment and --fail-on-no-slot 

    Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jul 19, 2026
    Configuration menu
    Copy the full SHA
    64d102d View commit details
    Browse the repository at this point in the history

Commits on Jul 21, 2026

  1. server: enforce -rea off by ignoring client reasoning overrides 

    When the server is started with -rea off (or --reasoning-format none),

do not allow API clients to re-enable reasoning via reasoning_format

or chat_template_kwargs.enable_thinking.

Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jul 21, 2026
    Configuration menu
    Copy the full SHA
    f1adb05 View commit details
    Browse the repository at this point in the history

  2. docs: document -rea off server-side enforcement 

    Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jul 21, 2026
    Configuration menu
    Copy the full SHA
    9f2eadf View commit details
    Browse the repository at this point in the history

Commits on Jul 27, 2026

  1. Merge remote-tracking branch 'upstream/master' into align-b10154

    @iz0eyj
    iz0eyj committed Jul 27, 2026
    Configuration menu
    Copy the full SHA
    83dd3dd View commit details
    Browse the repository at this point in the history

  2. llama.cpp-mv: update upstream tracker after b10154 alignment 

    Assisted-by: OpenCode
    @iz0eyj
    iz0eyj committed Jul 27, 2026
    Configuration menu
    Copy the full SHA
    6cb835a View commit details
    Browse the repository at this point in the history

Commits on Aug 1, 2026

  1. crow: ColQwen2 multivector vision — Qwen-VL smart_resize fix (PAD_NON… 

    …E + bicubic) + mtmd all-token embeddings
    @odellus
    odellus committed Aug 1, 2026
    Configuration menu
    Copy the full SHA
    46deb9f View commit details
    Browse the repository at this point in the history
Loading