Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 10, 2024

• whisper : document whisper_batch.n_seq_id (ggerganov#1942)

  Show description for a56f435
  josharian

  authored
  a56f435

  Copy full SHA for a56f435

• whisper : improve beam search candidate diversity (ggerganov#1947)

  Show description for ec16649
  josharian

  authored
  ec16649

  Copy full SHA for ec16649

Commits on Mar 9, 2024

• bindings/go : add linker flags to make metal work (ggerganov#1944)

  Show description for ccf022f
  josharian

  authored
  ccf022f

  Copy full SHA for ccf022f

• whisper : make beam candidate sort more stable (ggerganov#1943)

  Show description for 2852e1a
  josharian

  authored
  2852e1a

  Copy full SHA for 2852e1a

Commits on Mar 8, 2024

• ggml : try fix 32-bit arm compat (ggerganov#1938)

  Show description for ce945b5
  ggerganov

  authored
  ce945b5

  Copy full SHA for ce945b5

• talk-llama : use llama_decode instead of llama_eval

  ggerganov

  committed
  2f5a5a6

  Copy full SHA for 2f5a5a6

• talk-llama : sync llama.cpp

  ggerganov

  committed
  8e409d1

  Copy full SHA for 8e409d1

• talk-llama : sync llama.cpp

  ggerganov

  committed
  05d1b61

  Copy full SHA for 05d1b61

• sync : ggml

  ggerganov

  committed
  647cae1

  Copy full SHA for 647cae1

• Revert "[SYCL] fix error when set main gpu to non-zero (llama/5901)" (llama/5918)

  Show description for bae7c23
  

  NeoZhangJianyu

  authored and

  ggerganov

  committed
  bae7c23

  Copy full SHA for bae7c23

• fix error when set main gpu to non-zero (llama/5901)

  Show description for 18ea187
  

  NeoZhangJianyu

  authored and

  ggerganov

  committed
  18ea187

  Copy full SHA for 18ea187

• ggml : use SYS_get_cpu if SYS_getcpu is not defined (llama/5906)

  Show description for 1daeffc
  

  cebtenzzre

  authored and

  ggerganov

  committed
  1daeffc

  Copy full SHA for 1daeffc

• ggml : use uint8x16_t return type for ggml_vqtbl1q_u8 (llama/5894)

  Show description for 2f6f1d4
  

  bobqianic

  authored and

  ggerganov

  committed
  2f6f1d4

  Copy full SHA for 2f6f1d4

• add wait() to make code stable (llama/5895)

  

  NeoZhangJianyu

  authored and

  ggerganov

  committed
  7ff1894

  Copy full SHA for 7ff1894

• quants : use MM256_SET_M128I consistently to fix gcc 7 build (llama/5889)

  

  cebtenzzre

  authored and

  ggerganov

  committed
  8edfc54

  Copy full SHA for 8edfc54

• Vulkan Improvements (llama/5835)

  Show description for 9c39968
  

  0cc4m

  authored and

  ggerganov

  committed
  9c39968

  Copy full SHA for 9c39968

• fix mul_mat fault in CI/unit-test (llama/5862)

  Show description for 9d9a405
  

  21 peoplecommitted
  9d9a405

  Copy full SHA for 9d9a405

• ggml : fix unknown status (llama/0)

  ggerganov

  committed
  edd8b38

  Copy full SHA for edd8b38

• whisper : fix compute helper return (ggml/750)

  ggerganov

  committed
  ed76818

  Copy full SHA for ed76818

• ggml : introduce ggml_status (ggml/750)

  Show description for 9a0b59d
  

  3 peoplecommitted
  9a0b59d

  Copy full SHA for 9a0b59d

• cuda : fix data race in soft max (llama/5853)

  

  slaren

  authored and

  ggerganov

  committed
  93a84a1

  Copy full SHA for 93a84a1

• ggml : fix IQ3_S AVX implementation (llama/5834)

  Show description for bd26876
  ggerganov

  committed
  bd26876

  Copy full SHA for bd26876

• ggml : IQ3_S improvements (llama/5829)

  Show description for 21d2951
  

  2 peopleauthored and

  ggerganov

  committed
  21d2951

  Copy full SHA for 21d2951

• Support multiple GPUs (split mode) on SYCL backend (llama/5806)

  Show description for c3bfc9b
  

  NeoZhangJianyu

  authored and

  ggerganov

  committed
  c3bfc9b

  Copy full SHA for c3bfc9b

• ggml-vulkan: fix VULKAN_CHECK_RESULTS flag, which was previously broken (llama/5813)

  

  ddpasa

  authored and

  ggerganov

  committed
  422a6b1

  Copy full SHA for 422a6b1

• Use batched mul_mat pathway (llama/5591)

  Show description for 11dd0d4
  

  2 peopleauthored and

  ggerganov

  committed
  11dd0d4

  Copy full SHA for 11dd0d4

• make portability_enumeration_ext apple only (llama/5757)

  

  netrunnereve

  authored and

  ggerganov

  committed
  26dd2f0

  Copy full SHA for 26dd2f0

• add some new ops, fix some operators and add batch operations to certain operators. (ggml/747)

  Show description for 8cee7c0
  

  3 peoplecommitted
  8cee7c0

  Copy full SHA for 8cee7c0

Commits on Mar 6, 2024

• examples : Auto lowercase language parameter in main.cpp (ggerganov#1928)

  Show description for 2e2626b
  

  F1L1Pv2

  and

  bobqianic

  authored
  2e2626b

  Copy full SHA for 2e2626b

• examples : fix typo in bench.cpp (ggerganov#1933)

  zhouwg

  authored
  c0c0ae2

  Copy full SHA for c0c0ae2

Commits on Mar 5, 2024

• whisper : fix typo (ggerganov#1925)

  zhouwg

  authored
  897412b

  Copy full SHA for 897412b

• whisper.android.java : fix returns in JNI (ggerganov#1929)

  zhouwg

  authored
  f22d27a

  Copy full SHA for f22d27a

Commits on Mar 4, 2024

• cmake : add library versioning (ggerganov#1352)

  Show description for ccd7c1d
  

  kenneth-ge

  and

  ggerganov

  authored
  ccd7c1d

  Copy full SHA for ccd7c1d

• readme : recommend MacOS Sonoma for Core ML (ggerganov#1917)

  gavin1818

  authored
  c713eb5

  Copy full SHA for c713eb5

Commits on Feb 28, 2024

• talk-llama : sync llama.cpp

  ggerganov

  committed
  25d313b

  Copy full SHA for 25d313b

Pagination

Previous

Next