Commit Graph

  • 06d4fba851
    openai: align chat temperature and frequency_penalty options with completion (#6688) v0.3.10 frob 2024-09-07 18:08:08 +02:00
  • 70f849e4a3 Align Chat option processing with Completion option processing Richard Lyons 2024-09-07 16:16:23 +02:00
  • 9ba93168e7 Align Chat option processing with Completion option processing Richard Lyons 2024-09-07 16:02:21 +02:00
  • e9b7bb7864
    Merge branch 'ollama:main' into presence_penalty frob 2024-09-07 15:12:03 +02:00
  • 0cfc35140d
    Merge branch 'ollama:main' into cors frob 2024-09-07 15:11:46 +02:00
  • 108fb6c1d1
    docs: improve linux install documentation (#6683) Jeffrey Morgan 2024-09-06 22:05:37 -07:00
  • 09d1321489 docs: improve linux install documentation jmorganca 2024-09-06 22:01:15 -07:00
  • da915345d1
    openai: don't scale temperature or frequency_penalty (#6514) Yaroslav 2024-09-07 02:45:45 +02:00
  • 1012d06919
    Merge pull request #6682 from ollama/jessegross/goserver_logging Jesse Gross 2024-09-06 17:05:13 -07:00
  • 6ef4dd4921 llm: Remove debug logging Jesse Gross 2024-09-06 16:32:51 -07:00
  • 5b73fb33f5 llama.go: Set NDEBUG for CUDA Jesse Gross 2024-09-06 16:22:33 -07:00
  • 8a027bc401
    readme: add Archyve to community integrations (#6680) nickthecook 2024-09-06 17:06:01 -04:00
  • 5446903fbd
    readme: add Plasmoid Ollama Control to community integrations (#6681) imoize 2024-09-07 04:04:12 +07:00
  • a3b8793d0a
    readme: add Plasmoid Ollama Control to community integrations imoize 2024-09-07 03:02:31 +07:00
  • f851a82469 runner.go: Vision models are not yet supported Jesse Gross 2024-09-04 11:12:49 -07:00
  • 68e705fadf llama.go: Pass through additional context params Jesse Gross 2024-09-04 15:56:55 -07:00
  • 2f4920ecde llama: sync llama.cpp to commit 8962422 (#6621) Jeffrey Morgan 2024-09-04 15:14:49 -04:00
  • 0afde856e5 runner.go: Improve health status reporting when laoding model Jesse Gross 2024-09-03 12:05:46 -07:00
  • abf1288454 runner.go: Break out of loops for final tokens Jesse Gross 2024-09-03 10:25:39 -07:00
  • afed68a3e6 runner.go: Support GGUF LoRAs Jesse Gross 2024-08-28 17:12:06 -07:00
  • 5e8a0a585c runner.go: Don't cast a Go handle to a C void * Jesse Gross 2024-08-28 21:07:16 -07:00
  • 8a3a440ab2 runner.go: Support resource usage command line options Jesse Gross 2024-08-28 09:29:09 -07:00
  • 7651fb5b84 llama: fix sync script ggml-metal_darwin_arm64.m filename (#6610) Jeffrey Morgan 2024-09-03 14:01:52 -04:00
  • cdbd832b2d llama: delete unused files (#6523) Jeffrey Morgan 2024-08-29 17:30:11 -07:00
  • 479ed5bd18 runner.go: Update TODOs Jesse Gross 2024-08-27 17:05:23 -07:00
  • 6d59c0c0ba runner.go: Fix embeddings endpoint Jesse Gross 2024-08-27 13:59:33 -07:00
  • 3abce86015 runner.go: Health endpoint comments Jesse Gross 2024-08-27 13:17:04 -07:00
  • 852f8af8d3 runner.go: Cleanups Jesse Gross 2024-08-27 13:19:46 -07:00
  • d1806db773 runner.go: Move pieces[] into sequence Jesse Gross 2024-08-27 10:24:33 -07:00
  • bfaf75bbf8 runner.go: Fix deadlock if a connection is closed during decoding Jesse Gross 2024-08-26 14:20:50 -07:00
  • 147200a324 runner.go: Fix resource leaks when removing sequences Jesse Gross 2024-08-26 14:26:48 -07:00
  • aed338f0f9 runner.go: Separate KV cache and context sizes Jesse Gross 2024-08-23 17:27:09 -07:00
  • 24357b00b4 runner.go: Hold mutex for entire time when processing batch Jesse Gross 2024-08-23 16:28:38 -07:00
  • 995fc43995 runner.go: Scale batches to be processed by numParallel Jesse Gross 2024-08-23 13:44:30 -07:00
  • a90dad733a Make new tokenizer logic conditional (#6395) Daniel Hiltgen 2024-08-24 17:25:37 -07:00
  • 5357594a6e runner.go: Support MinP parameter Jesse Gross 2024-08-21 16:13:54 -07:00
  • 9f291c1f8b runner.go: Check for incomplete UTF-8 character Jesse Gross 2024-08-15 13:07:28 -07:00
  • 29161195ec runner.go: Implement RepeatLastN to penalize repeated tokens Jesse Gross 2024-08-20 11:21:19 -07:00
  • a5330e655e runner.go: Use correct JSON field names for runners Jesse Gross 2024-08-20 16:58:09 -07:00
  • 0e6b64c1dd runner.go: Shift context window when KV cache space is exceeded Jesse Gross 2024-08-14 10:35:49 -07:00
  • e3ba02887b runner.go: Don't decode if nothing has been added to the batch Jesse Gross 2024-08-14 10:32:05 -07:00
  • 39c5c9976b llama.go: Advance though tokens when processing multiple batches Jesse Gross 2024-08-13 16:53:35 -07:00
  • 37eb39e33f llama.go: Use dynamic buffer for TokenToPiece Jesse Gross 2024-08-19 17:54:57 -07:00
  • 4301aaf8d6 llama.go: Make batch memory allocation match configuration Jesse Gross 2024-08-13 11:18:02 -07:00
  • 50a6267dd1 runner.go: Fix off by one in batch size check Jesse Gross 2024-08-13 10:51:50 -07:00
  • 6e4550e7a7 llm: Fix array out-of-bounds memory access when tokenizing Jesse Gross 2024-08-15 15:25:21 -07:00
  • 3f602270c0 runner: Initialize numPredict Jesse Gross 2024-08-13 10:38:03 -07:00
  • a8892c1c01 server: Fix double free on runner subprocess error. Jesse Gross 2024-08-16 14:46:33 -07:00
  • 0b86e359ba llm: Fix lint Jesse Gross 2024-08-15 09:50:42 -07:00
  • 78dab063d3 Fix filename for non darwin arm builds Daniel Hiltgen 2024-08-13 14:33:11 -07:00
  • d13c136186 lint jmorganca 2024-08-13 11:24:46 -07:00
  • 6b1bf96f98 Add missing vendor headers to ggml sync Daniel Hiltgen 2024-08-09 16:45:18 -07:00
  • 0f51984487 Wire up native source file dependencies Daniel Hiltgen 2024-08-05 08:56:47 -07:00
  • 1f1918243f Bump llama sync to 1e6f65 Daniel Hiltgen 2024-08-06 16:50:34 -07:00
  • 9d7693e5eb fix dolphin-mistral Daniel Hiltgen 2024-08-01 14:47:00 -07:00
  • 78697ea0a9 harden integration tests Daniel Hiltgen 2024-08-01 14:41:23 -07:00
  • d4f604ed0d Runtime selection of new or old runners Daniel Hiltgen 2024-08-01 08:54:44 -07:00
  • 10bf065fda Implement timings response in Go server Daniel Hiltgen 2024-07-29 14:09:55 -07:00
  • ac36093015 Get embeddings working Daniel Hiltgen 2024-07-31 11:08:09 -07:00
  • 1406c068cd Fix parallel requests Daniel Hiltgen 2024-07-31 15:02:58 -07:00
  • 4788212f03 Update sync with latest llama.cpp layout, and run against b3485 Daniel Hiltgen 2024-07-29 16:21:09 -07:00
  • ff8a3c4d30 Prefix all build artifacts with an OS/ARCH dir Daniel Hiltgen 2024-06-24 09:23:34 -07:00
  • 8e49d3b3ab Get linux building Daniel Hiltgen 2024-06-23 12:07:41 -07:00
  • 042ad3faba add note in readme jmorganca 2024-06-21 16:22:27 -04:00
  • 31f0d9c5ba clean up metal code jmorganca 2024-06-15 10:06:36 -07:00
  • 402348dde8 fix Makefile on windows jmorganca 2024-06-20 21:52:10 -04:00
  • a5664b7e13 remove printing jmorganca 2024-06-13 18:41:12 -07:00
  • 064f49a430 dont apply license to stb_image.h and json.hpp jmorganca 2024-06-13 14:35:11 -07:00
  • d1932a2a19 lint jmorganca 2024-06-13 14:21:55 -07:00
  • c00e353d87 update sync header jmorganca 2024-06-13 14:12:23 -07:00
  • 50fbdcb5b6 remove unused script jmorganca 2024-06-13 14:07:05 -07:00
  • e95b24e008 fix metal jmorganca 2024-06-12 12:18:40 -07:00
  • af958935df add header to not edit jmorganca 2024-06-12 11:40:13 -07:00
  • af3049c34f add header to not edit jmorganca 2024-06-12 11:38:42 -07:00
  • 71049338b3 fix build on windows jmorganca 2024-06-12 02:47:12 -04:00
  • 485a29604d fix Makefile jmorganca 2024-06-11 23:18:07 -07:00
  • c2c2bc4898 fix README.md jmorganca 2024-06-11 22:54:45 -07:00
  • d6f351ac3b fix README.md jmorganca 2024-06-11 22:54:31 -07:00
  • 11a67b04dc consistent whitespace jmorganca 2024-06-11 22:50:10 -07:00
  • 20a206a961 update .gitattributes jmorganca 2024-06-11 22:48:06 -07:00
  • f771b1ae15 link metal jmorganca 2024-06-11 22:46:14 -07:00
  • 594e1422f3 wip jmorganca 2024-06-11 18:53:48 -07:00
  • ecaf1efc9e wip meta jmorganca 2024-06-11 11:12:00 -07:00
  • b76839a212 sync jmorganca 2024-06-10 17:23:09 -07:00
  • 01539da2b5 remove perl docs jmorganca 2024-06-10 09:26:19 -07:00
  • b68f58426a remove build scripts jmorganca 2024-06-10 02:56:37 -04:00
  • ff10679b0b remove need for perl jmorganca 2024-06-10 00:04:21 -04:00
  • 7908ba18d2 fix output jmorganca 2024-06-09 23:53:40 -04:00
  • a94126b7f8 arch build jmorganca 2024-06-09 20:19:11 -07:00
  • 9dca026c34 add temporary makefile jmorganca 2024-06-09 22:33:31 -04:00
  • ce029c30a4 fix cuda and rocm builds jmorganca 2024-06-09 19:49:22 -04:00
  • ac69660897 fix cgo flags for darwin amd64 jmorganca 2024-06-09 14:30:41 -07:00
  • 697beec40d remove -fPIC from build_hipblas.sh jmorganca 2024-06-07 12:52:49 -04:00
  • 61fb941001 fix issues with runner jmorganca 2024-06-07 09:32:52 -07:00
  • bf8d7950ba move sync script back in for now jmorganca 2024-06-07 09:26:44 -07:00
  • b4df0d717b llama: sync jmorganca 2024-06-07 00:27:24 -07:00
  • 78f0654cbf update to d5c938cd jmorganca 2024-06-07 00:15:58 -07:00
  • 8fcec9258b add patches jmorganca 2024-06-06 23:55:47 -07:00
  • 1537929ee9 cleanup stop code jmorganca 2024-06-04 00:58:58 -07:00
  • 1603373fc8 fix example jmorganca 2024-06-04 00:43:03 -07:00