record(ORACLE-LLAMACPP-REPIN-STOCK): the pin named a tree that exists nowhere (#857) by localai-bot · Pull Request #1051 · mudler/vllm.cpp

spec(ORACLE-LLAMACPP-REPIN-STOCK): the llama.cpp floor was our own fork, 65 perf commits deep (#1003, #857)

mudler

committed

38a718a

record(ORACLE-LLAMACPP-REPIN-STOCK): repin llama.cpp to stock b10451, gateable=no (#857, #1003)

mudler

committed

a2ede63

merge: origin/main into the ORACLE-LLAMACPP-REPIN-STOCK repin

mudler

committed

3f63066

record(ORACLE-LLAMACPP-REPIN-STOCK): the repin's own claims, re-derived (#1003, #857)

mudler

committed

24e47f2

merge: origin/main into the ORACLE-LLAMACPP-REPIN-STOCK re-derivation

mudler

committed

fa79980

record(ORACLE-LLAMACPP-REPIN-STOCK): the gate section said two reds, the tree now has three others (#1003)

mudler

committed

bab8e1f

record(ORACLE-LLAMACPP-REPIN-STOCK): derive the contaminated set instead of listing it, and the sweep finds a fifth win, a third revision and a shallow clone (#1003, #857)

mudler

committed

489e93f

merge: origin/main into the ORACLE-LLAMACPP-REPIN-STOCK sweep, keyed record re-applied by hand

mudler

committed

e34e3e4

record(ORACLE-LLAMACPP-REPIN-STOCK): the gate section is green now, and the anchor count moved a third time without an edit (#1003, #857)

mudler

committed

bf62128

record(ORACLE-LLAMACPP-REPIN-STOCK): the sweep's own path set was a hand list, and README.md fell outside it (#1003, #857)

mudler

committed

5de5236

record(ORACLE-LLAMACPP-REPIN-STOCK): mark the front page, and the one denominator that reaches a shipped default (#1003)

mudler

committed

85a9a7a

record(ORACLE-LLAMACPP-REPIN-STOCK): the × token was present and dead, and it hid a fifth llama.cpp revision (#1003, #857)

mudler

committed

8e0bf0e

merge: origin/main into the ORACLE-LLAMACPP-REPIN-STOCK review repairs

mudler

committed

b60ad20

record(ORACLE-LLAMACPP-REPIN-STOCK): the gate block headings name their SHA now, and the flake gets a control pair (#1003, #618)

mudler

committed

7395b77

record(ORACLE-LLAMACPP-REPIN-STOCK): the same head is green at load 15 and red at load 65 (#618, #1003)

mudler

committed

2f0cd62

record(ORACLE-LLAMACPP-REPIN-STOCK): the self-test covered 6 of 18 tokens, and stage 2 was a second spelling of stage 1 (#1003)

mudler

committed

fa94b10

record(ORACLE-LLAMACPP-REPIN-STOCK): the self-test proved liveness, not shape coverage, and the attribution half had no control (#1003)

mudler

committed

adcf2bb

merge: origin/main into the #857 landing

mudler

committed

cb07436

Navigation Menu

record(ORACLE-LLAMACPP-REPIN-STOCK): the pin named a tree that exists nowhere (#857) - #1051

record(ORACLE-LLAMACPP-REPIN-STOCK): the pin named a tree that exists nowhere (#857)#1051
localai-bot merged 18 commits into
mainfrom
row/ORACLE-LLAMACPP-REPIN-STOCK

Commits on Aug 16, 2026

spec(ORACLE-LLAMACPP-REPIN-STOCK): the llama.cpp floor was our own fork, 65 perf commits deep (#1003, #857)

record(ORACLE-LLAMACPP-REPIN-STOCK): repin llama.cpp to stock b10451, gateable=no (#857, #1003)

merge: origin/main into the ORACLE-LLAMACPP-REPIN-STOCK repin

record(ORACLE-LLAMACPP-REPIN-STOCK): the repin's own claims, re-derived (#1003, #857)

merge: origin/main into the ORACLE-LLAMACPP-REPIN-STOCK re-derivation

record(ORACLE-LLAMACPP-REPIN-STOCK): the gate section said two reds, the tree now has three others (#1003)

record(ORACLE-LLAMACPP-REPIN-STOCK): derive the contaminated set instead of listing it, and the sweep finds a fifth win, a third revision and a shallow clone (#1003, #857)

merge: origin/main into the ORACLE-LLAMACPP-REPIN-STOCK sweep, keyed record re-applied by hand

record(ORACLE-LLAMACPP-REPIN-STOCK): the gate section is green now, and the anchor count moved a third time without an edit (#1003, #857)

record(ORACLE-LLAMACPP-REPIN-STOCK): the sweep's own path set was a hand list, and README.md fell outside it (#1003, #857)

record(ORACLE-LLAMACPP-REPIN-STOCK): mark the front page, and the one denominator that reaches a shipped default (#1003)

record(ORACLE-LLAMACPP-REPIN-STOCK): the × token was present and dead, and it hid a fifth llama.cpp revision (#1003, #857)

merge: origin/main into the ORACLE-LLAMACPP-REPIN-STOCK review repairs

record(ORACLE-LLAMACPP-REPIN-STOCK): the gate block headings name their SHA now, and the flake gets a control pair (#1003, #618)

record(ORACLE-LLAMACPP-REPIN-STOCK): the same head is green at load 15 and red at load 65 (#618, #1003)

record(ORACLE-LLAMACPP-REPIN-STOCK): the self-test covered 6 of 18 tokens, and stage 2 was a second spelling of stage 1 (#1003)

record(ORACLE-LLAMACPP-REPIN-STOCK): the self-test proved liveness, not shape coverage, and the attribution half had no control (#1003)

merge: origin/main into the #857 landing