FazBrowse GitHub Viewer | Trending |
URL:
| Home
Tools: [Download Repo ZIP]   [Original HTTPS Page]

Comparing craftingmod:main...JamePeng:main · craftingmod/llama-cpp-python · GitHub

/ llama-cpp-python Public
forked from JamePeng/llama-cpp-python
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: craftingmod/llama-cpp-python
Failed to load repositories. Confirm that selected base ref is valid, then try again.
Loading
base: main
Choose a base ref
Could not load branches
Nothing to show
{{ refName }}
...
head repository: JamePeng/llama-cpp-python
Failed to load repositories. Confirm that selected head ref is valid, then try again.
Loading
compare: main
Choose a head ref
Could not load branches
Nothing to show
{{ refName }}
Checking mergeability… Don’t worry, you can still create the pull request.
  • 5 commits
  • 6 files changed
  • 1 contributor

Commits on Aug 24, 2026

  1. Update Submodule vendor/llama.cpp c060ca9..f280b26 

    Signed-off-by: JamePeng <jame_peng@sina.com>
    JamePeng committed Aug 24, 2026
    Configuration menu
    Copy the full SHA
    538a45e View commit details
    Browse the repository at this point in the history

Commits on Aug 25, 2026

  1. Update Submodule vendor/llama.cpp f280b26..eab8ee4 

    Signed-off-by: JamePeng <jame_peng@sina.com>
    JamePeng committed Aug 25, 2026
    Configuration menu
    Copy the full SHA
    578037d View commit details
    Browse the repository at this point in the history

  2. fix(cache): track native positions for hybrid checkpoint rollback 

    - store native memory pos_min and pos_max in HybridCheckpoint
- remove the attention-memory suffix from pos_max + 1 after partial restore
- preserve compatibility with legacy checkpoints without native positions
- return failure when suffix cleanup cannot be completed
- add checkpoint position and rollback cleanup tests
- update HybridCheckpointCache documentation

Signed-off-by: JamePeng <jame_peng@sina.com>
    JamePeng committed Aug 25, 2026
    Configuration menu
    Copy the full SHA
    dc24f2e View commit details
    Browse the repository at this point in the history

Commits on Aug 26, 2026

  1. Update Submodule vendor/llama.cpp eab8ee4..4d19b28 

    Signed-off-by: JamePeng <jame_peng@sina.com>
    JamePeng committed Aug 26, 2026
    Configuration menu
    Copy the full SHA
    abab290 View commit details
    Browse the repository at this point in the history

  2. fix(speculative): align MTP context and verification limits 

    - size draft contexts from the initialized target context n_ctx
- warn when MTP prompt processing leaves the draft cache incomplete
- skip prompt-cache diagnostics for shared-memory MTP contexts
- cap draft proposals at n_batch - 1 to keep verification batches atomic
- preserve the user-requested draft_n_max while applying the runtime limit

Signed-off-by: JamePeng <jame_peng@sina.com>
    JamePeng committed Aug 26, 2026
    Configuration menu
    Copy the full SHA
    3d9bc6d View commit details
    Browse the repository at this point in the history
Loading
Back | FazBrowse Home | New Git URL