vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Commit graph

  • dc913ea3c4 minor M. Yusuf Sarıgöz 2023-10-12 10:48:11 +03:00
  • 0ee4ff4692 Merge branch 'master' of github.com:ggerganov/llama.cpp Laura 2023-10-12 09:31:22 +02:00
  • 8d1cd512e2 missed a flag Concedo 2023-10-12 15:00:51 +08:00
  • c6fe820357 improve cors and header handling Concedo 2023-10-12 14:53:39 +08:00
  • b016596d90
         server : add completion mode (no chat) (#3582) b1372 Aarni Koskela 2023-10-12 15:51:53 +09:00
  • 6b3ae4da92
         prompts : add mnemonics.txt Georgi Gerganov 2023-10-12 09:35:19 +03:00
  • 57dd55e2c7
         server : fix kv cache management (#3588) b1370 Georgi Gerganov 2023-10-12 09:29:04 +03:00
  • f604cffdce multiuser racer bugfix Concedo 2023-10-12 13:39:12 +08:00
  • e6ea63ca71 simply ignore added tokens that id < vocab size 김승덕/Infrastructure그룹(YA) 2023-10-12 12:44:33 +09:00
  • 228a0f8545
         typo: it is --n-gpu-layers not --gpu-layers Ian Scrivener 2023-10-12 14:34:40 +11:00
  • aa7fbef78c Refactor graph building to reduce duplication KerfuffleV2 2023-10-11 20:04:01 -06:00
  • 471230202d crash fixed FSSRepo 2023-10-11 19:48:15 -04:00
  • f7b1205a51 formatting, remove copying iterator on delete staviq 2023-10-12 00:26:44 +02:00
  • 63f99b1ea6 implementing parallel decoding in server example FSSRepo 2023-10-11 18:14:11 -04:00
  • a20b3e6cf0 fix missing import Cebtenzzre 2023-10-11 17:36:11 -04:00
  • 1403d87cca Merge master and fix conflicts M. Yusuf Sarıgöz 2023-10-12 00:00:57 +03:00
  • 058e83ca7b
         server : fix kv cache management Georgi Gerganov 2023-10-12 00:00:54 +03:00
  • b8fe4b5cc9
         main : fix session loading bug (#3400) b1369 Georgi Gerganov 2023-10-11 23:55:08 +03:00
  • eac5f5442c
         Merge branch 'master' into specialtokens staviq 2023-10-11 22:54:29 +02:00
  • 29e6b46e03 use offsets instead of copy by substr staviq 2023-10-11 22:47:17 +02:00
  • 2bc1710e2b command line: use gpt_params_parse() M. Yusuf Sarıgöz 2023-10-11 23:17:50 +03:00
  • b44722946e Fix wg_denom of m-mulmat shaders 0cc4m 2023-10-11 22:08:38 +02:00
  • a8bdd65525
         server : add parameter -tb N, --threads-batch N (#3584) b1368 Michael Coppola 2023-10-11 15:42:22 -04:00
  • 70c29da118
         common : fix mirostat state when using multiple sequences (#3543) b1367 Kerfuffle 2023-10-11 13:35:46 -06:00
  • 80b2e729fa Remove gpt neox references Galunid 2023-10-11 21:05:40 +02:00
  • 27467a5ced Initial support - model loads, generates random stuff Galunid 2023-10-11 20:37:22 +02:00
  • 8c70a5ff25
         batched : add bench tool (#3545) b1366 Georgi Gerganov 2023-10-11 21:25:33 +03:00
  • 576df7770a beautify 김승덕/Infrastructure그룹(YA) 2023-10-12 03:03:37 +09:00
  • 9b5907ead7 Override tokens if their ids in added tokens are below the vocab_base size 김승덕/Infrastructure그룹(YA) 2023-10-12 02:53:09 +09:00
  • fffa4c0099 Use more consistant naming convention for sampling contexts KerfuffleV2 2023-10-11 11:48:23 -06:00
  • e86a7d2ebd avoid 100% cpu usage all time FSSRepo 2023-10-11 13:03:35 -04:00
  • a003e3c348 horde auto recovery Concedo 2023-10-12 00:57:32 +08:00
  • 2fcdf869cd
         batched-bench : add mmq CLI arg batched-bench Georgi Gerganov 2023-10-11 19:42:33 +03:00
  • daeb834da9
         batched-bench : pass custom set of PP, TG and PL Georgi Gerganov 2023-10-11 19:36:31 +03:00
  • 0fd0f284be
         Merge branch 'ggerganov:master' into master Michael Coppola 2023-10-11 12:29:12 -04:00
  • c062ffd18c
         batched-bench : init warm-up batch Georgi Gerganov 2023-10-11 19:24:59 +03:00
  • 32b237a1e2 Merge branch 'master' of https://github.com/ggerganov/llama.cpp into fixes FSSRepo 2023-10-11 12:22:01 -04:00
  • 76e17f8d93
         Merge branch 'master' into batched-bench Georgi Gerganov 2023-10-11 19:18:35 +03:00
  • 026bb1b1cd
         batched-bench : add readme + n_kv_max is now configurable Georgi Gerganov 2023-10-11 19:09:50 +03:00
  • 87a0361eae server : add completion mode (no chat) Aarni Koskela 2023-10-10 14:21:56 +09:00
  • f0f78345f2 Use temperature = 0.1 by default M. Yusuf Sarıgöz 2023-10-11 15:03:01 +03:00
  • 24ba3d829e
         examples : add batched.swift + improve CI for swift (#3562) b1365 Zane Shannon 2023-10-11 04:14:05 -07:00
  • 4a34e63500 Apply some renaming suggestions. KerfuffleV2 2023-10-11 04:14:47 -06:00
  • 51e9d39117
         Update convert-baichuan-hf-to-gguf.py Georgi Gerganov 2023-10-11 09:55:15 +03:00
  • cb56fb1319 [examples/batched.swift] add an example swift implementation of batched & build it in ci to validate SPM is correctly configured for dependencies Zane Shannon 2023-10-10 22:48:11 -07:00
  • 0409ae00b6 are you happy editorconfig? M. Yusuf Sarıgöz 2023-10-11 08:21:29 +03:00
  • ab2158796f Check if apples are compared to apples M. Yusuf Sarıgöz 2023-10-11 08:15:51 +03:00
  • f1564bb2eb Merge branch 'master' into llava M. Yusuf Sarıgöz 2023-10-11 06:59:37 +03:00
  • 587bde8e0c Maybe seed is unlucky? M. Yusuf Sarıgöz 2023-10-11 06:40:52 +03:00
  • 9f6ede19f3
         Add MPT model to supported models in README.md (#3574) Galunid 2023-10-11 01:02:49 +02:00
  • 1e506b5c75 Add MPT model to supported models in README.md Galunid 2023-10-11 00:40:43 +02:00
  • 754240e7bf
         Merge 4b8d5e3890 into 233fc1c69f Georgi Gerganov 2023-10-10 20:49:01 +00:00
  • 128562dc83
         Create devcontainer.json kaaid 2023-10-10 22:31:48 +03:00
  • 233fc1c69f
         Minor improvements in GPT2 tokenizer (#3567) b1363 goerch 2023-10-10 18:59:52 +02:00
  • aebe027100 fixing potential NULL dereference in GGML allocator tpltnt 2023-10-10 18:47:37 +02:00
  • 7e60415bc9 fixing potential NULL dereference in reading GGUF file tpltnt 2023-10-10 18:27:00 +02:00
  • c5b49360d0
         readme : add bloom (#3570) Xingchen Song(宋星辰) 2023-10-11 00:28:50 +08:00
  • 94496cd49c fixing potential NULL dereference in training tpltnt 2023-10-10 18:19:01 +02:00
  • c5d2415e6e fixing tiny resource leak when reading /proc/cpuinfo on Linux tpltnt 2023-10-10 18:00:10 +02:00
  • 515daff06f readme: add bloom xingchensong 2023-10-10 23:09:22 +08:00
  • 02d2875def
         llm : add bloom models (#3553) Xingchen Song(宋星辰) 2023-10-10 22:48:21 +08:00
  • fc634d87a8 shorten param name, add st verification by type staviq 2023-10-10 16:34:24 +02:00
  • 9c62605f3c
         Merge branch 'master' into HEAD Georgi Gerganov 2023-10-10 17:09:44 +03:00
  • 37ae96c712 Correct bug pudepiedj 2023-10-10 13:50:28 +01:00
  • 37050ba90a Updated cmap-example pudepiedj 2023-10-10 13:33:12 +01:00
  • 0aa6595ae0
         swift : improvements and fixes (#3564) b1360 Jhen-Jie Hong 2023-10-10 06:31:13 -05:00
  • d74eab0e63 actually for this round, do not include deprecated params. i dont want to have to deal with them (+2 squashed commit) Concedo 2023-10-10 18:51:04 +08:00
  • d640aae755 add support for 13b model variant M. Yusuf Sarıgöz 2023-10-10 13:02:24 +03:00
  • a723466d50 Merge branch 'master' into concedo_experimental Concedo 2023-10-10 17:21:42 +08:00
  • 1b25b21655
         Merge pull request #27 from one-lithe-rune/allow-sdk-dll-loading - Allow use of hip SDK (if installed) dlls on windows (#470) YellowRoseCx 2023-10-10 04:16:33 -05:00
  • 9b608da8ad
         Merge branch 'ggerganov:master' into master vvhg1 2023-10-10 09:56:51 +02:00
  • f5f9121de1
         llm : add MPT support (#3417) b1359 Jan Ploski 2023-10-10 09:50:23 +02:00
  • 9b66378cac
         llama : print clam_kqv and max_alibi_bias hparams Georgi Gerganov 2023-10-10 10:49:39 +03:00
  • 96cf3f5dc3
         ggml : remove obsolete n_past assert in ggml_alibi Georgi Gerganov 2023-10-10 10:45:24 +03:00
  • 11ea5c7d96
         infill. : fix tokenization (#3508) b1358 vvhg1 2023-10-10 09:31:21 +02:00
  • d0c8d14eae Don't add bos token in test goerch 2023-10-10 07:41:49 +02:00
  • 8d0c57583c Fixing minor bugs in bpe_gpt2_preprocess goerch 2023-10-10 07:32:03 +02:00
  • 899cb6583f swift : add -O3 -DNDEBUG unsafe flags Jhen 2023-10-10 08:42:28 +08:00
  • 1558b90d28 swift : add missing ggml-backend.c source Jhen 2023-10-10 08:42:10 +08:00
  • 606c384114 swift : use macOS 12 as minimum requirement Jhen 2023-10-10 08:37:03 +08:00
  • 96171de5ef add llava target to Makefile M. Yusuf Sarıgöz 2023-10-10 01:50:02 +03:00
  • 5009ae90ef Handle cases where image file does not exist M. Yusuf Sarıgöz 2023-10-10 01:49:35 +03:00
  • ae01c859e5 gitignore /llava M. Yusuf Sarıgöz 2023-10-10 01:13:12 +03:00
  • 759973be79 printf format pudepiedj 2023-10-09 23:11:43 +01:00
  • 301f13cf9b Compile new cmap-example pudepiedj 2023-10-09 22:49:57 +01:00
  • bddd099edf experimental pudepiedj 2023-10-09 22:17:17 +01:00
  • d75a0315f0 are you happy editorconfig? M. Yusuf Sarıgöz 2023-10-09 23:56:07 +03:00
  • 325d240061 introduce pad-to-square mode for non-square images M. Yusuf Sarıgöz 2023-10-09 23:53:29 +03:00
  • 47675b9200 experiment in C pudepiedj 2023-10-09 21:00:49 +01:00
  • ba32402291 creadcommonh pudepiedj 2023-10-09 20:38:00 +01:00
  • e2962e1262 Minor fixes 0cc4m 2023-10-09 21:15:44 +02:00
  • 351772912d Merge branch 'master' of github.com:ggerganov/llama.cpp vvhg1 2023-10-09 20:51:37 +02:00
  • f288c6b5e3 Merge branch 'master' into concedo_experimental Concedo 2023-10-10 00:09:46 +08:00
  • 53dbefa2f8 Added final newline (2) pudepiedj 2023-10-09 17:05:24 +01:00
  • 6189a9ef3a One more trailing ws pudepiedj 2023-10-09 16:55:31 +01:00
  • 9abc92545c Remove trailing ws pudepiedj 2023-10-09 16:53:27 +01:00
  • 32172f12f5 mpt : remove unused tokenizer_json in convert script Cebtenzzre 2023-10-09 11:39:53 -04:00
  • 96e9539f05
         OpenAI compat API adapter (#466) Matěj Štágl 2023-10-09 17:24:48 +02:00
  • 7636c34891 Merge branch 'context-sensitive-help' of https://github.com/pudepiedj/llama.cpp into context-sensitive-help pudepiedj 2023-10-09 16:10:15 +01:00
  • 094d6d6e09 Add help list pudepiedj 2023-10-09 16:10:10 +01:00