Commit graph

  • 65d9f7349d
    add ggml_log operation necessary for cross entropy loss xaedes 2023-05-06 17:35:13 +02:00
  • 8cf04fec9d
    fix soft_max backward pass for input->ne[1] != 1 xaedes 2023-05-06 17:30:38 +02:00
  • b4c273f7a3
    add ggml_reshape_1d, ggml_reshape_4d and ggml_view_4d xaedes 2023-05-06 17:29:41 +02:00
  • 2986951203
    fix Henri Vasserman 2023-05-06 18:02:05 +03:00
  • 0dfa17db1e
    fix Henri Vasserman 2023-05-06 17:59:44 +03:00
  • 87d8ac9286
    fix Henri Vasserman 2023-05-06 17:59:26 +03:00
  • 5cb13c2fcc
    add version numbers Henri Vasserman 2023-05-06 17:57:02 +03:00
  • 71fac5bbcb
    Merge 'origin/master' into cistuff Henri Vasserman 2023-05-06 16:57:21 +03:00
  • 46879f35a6
    Remove default arguments from sampling functions Jed Fox 2023-05-06 09:25:43 -04:00
  • 5d4640511b
    Add *.bin to gitignore Jed Fox 2023-05-06 09:25:02 -04:00
  • f1d51d144b
    train on multiple examples, generate & print tokens with trained model afterwards xaedes 2023-05-06 14:13:55 +02:00
  • 83ee1cd741
    fix bug when using ggml_opt to optimize params in one context and use a renewable context for eval and opt xaedes 2023-05-06 13:05:29 +02:00
  • 23f148516d use _mm_pause() in busyloop Sami Farin 2023-05-06 07:48:45 +03:00
  • ff93b394da fixed a typo Concedo 2023-05-06 12:37:34 +08:00
  • a48dddab86 slightly bump the RAM up to support chinese alpaca Concedo 2023-05-06 11:48:22 +08:00
  • 1fa3128d2a
    Allow usage of CLBlast alongside Accelerate.framework swittk 2023-05-06 08:24:48 +07:00
  • fe72648fac feat: add alibi Zach Nussbaum 2023-05-05 20:21:04 -04:00
  • 173d0e6419
    makefile: automatic Arch Linux detection (#1332) master-173d0e6 DaniAndTheWeb 2023-05-05 23:57:14 +02:00
  • 1107194e6b
    Merge 'origin/master' into hipblas Henri Vasserman 2023-05-06 00:38:20 +03:00
  • a44c384c56
    Shrink Arch Linux check DaniAndTheWeb 2023-05-05 23:23:47 +02:00
  • ab5e61d710
    Fix possible error DaniAndTheWeb 2023-05-05 23:10:28 +02:00
  • a3b85b28da
    ci : add cublas to windows release (#1271) master-a3b85b2 Erik Scholz 2023-05-05 22:56:09 +02:00
  • 471624b464
    Rename variable DaniAndTheWeb 2023-05-05 22:52:49 +02:00
  • c0a86cc556
    Optimize Arch Linux check DaniAndTheWeb 2023-05-05 22:51:56 +02:00
  • e0ed30dffe Dockerfiles: use LLAMA_AVX2_ONLY. Bryan Silverthorn 2023-05-05 10:49:03 -07:00
  • 2e3b2e7e09 Makefile: support LLAMA_AVX2_ONLY. Bryan Silverthorn 2023-05-05 10:48:13 -07:00
  • 8a964e76c8 integrated mirostat as a launch parameter, works on all models Concedo 2023-05-06 00:47:17 +08:00
  • a8815a683e
    Remove Arch Linux note DaniAndTheWeb 2023-05-05 18:31:48 +02:00
  • 4e3c178be4
    Automatic Arch Linux detection for BLAS DaniAndTheWeb 2023-05-05 18:31:05 +02:00
  • 851f55325a Merge remote-tracking branch 'temp/concedo' into concedo_experimental Concedo 2023-05-05 23:55:53 +08:00
  • 534c89e766 Track character width Danny Daemonic 2023-05-05 08:38:58 -07:00
  • 921dcee00a
    readme: add missing info (#1324) Pavol Rusnak 2023-05-05 16:43:36 +02:00
  • 60196ae73d ggml: add AVX support katsu560 2023-05-05 23:39:07 +09:00
  • 2edbcebe27 added optional force versioning flag Concedo 2023-05-05 22:02:00 +08:00
  • 39f3d1cf48 Merge branch 'master' into concedo_experimental Concedo 2023-05-05 21:34:33 +08:00
  • 2d13786e91
    Fix for OpenCL / clbast builds on macOS. (#1329) master-2d13786 Ionoclast Laboratories 2023-05-05 08:18:21 -04:00
  • 8f9f962d4d Signed variable to unsigned variable cast Danny Daemonic 2023-04-29 17:48:52 -07:00
  • 94dd17247a author mode -> multiline input Danny Daemonic 2023-04-23 07:05:02 -07:00
  • 52e319050b Add author mode and other related QOL improvements Danny Daemonic 2023-04-18 02:55:40 -07:00
  • 8131bc8b56 add new sampling algorithm mirostat Hendrik Langer 2023-05-05 13:23:47 +02:00
  • 46da5de195
    Revert "quick readme update" CRD716 2023-05-04 21:30:51 -05:00
  • a90e96b266
    Convert.py @staticmethod (#1327) Benjamin Lecaillon 2023-05-05 02:17:07 +02:00
  • b05ec02a1b
    Update convert.py Ivan Stepanov 2023-05-05 03:06:26 +03:00
  • 100fc2be5e
    Fix for OpenCL / clbast builds on macOS. Ionoclast Laboratories 2023-05-04 19:09:45 -04:00
  • 94c5652fc0
    quantize: make output filename optional, default to ggml-model-<ftype>.bin (#1301) master-94c5652 slaren 2023-05-05 00:58:56 +02:00
  • 893bf9d368
    Line 698 has one #staticmethod and should not Benjamin Lecaillon 2023-05-04 23:49:01 +02:00
  • 47bbd631f2
    readme: add missing info Pavol Rusnak 2023-05-04 20:59:19 +02:00
  • 34d9f22f44
    Wrap exceptions in std::exception to verbose output on exception. (#1316) master-34d9f22 Ivan Stepanov 2023-05-04 19:56:27 +03:00
  • d3e8093e9b
    convert: support DT_BF16 tensors (#1309) Ivan Stepanov 2023-05-04 19:54:37 +03:00
  • 360cfe5bec
    readme : add OpenBuddy link (#1321) 44670 2023-05-05 00:33:31 +08:00
  • 92e2b38a9a
    more jank Henri Vasserman 2023-05-04 19:26:45 +03:00
  • 52179eb4d9
    MSVC stuff Henri Vasserman 2023-05-04 19:05:43 +03:00
  • ae28ec9429
    Update README.md 44670 2023-05-05 00:18:43 +08:00
  • 2edbdb0f99
    main : add --in-suffix option (#1318) master-2edbdb0 44670 2023-05-04 23:41:12 +08:00
  • 07b8ddb743
    Merge 'origin/master' into cistuff Henri Vasserman 2023-05-04 18:31:08 +03:00
  • b0d9e4c322
    not sure why this is failing Henri Vasserman 2023-05-04 18:22:03 +03:00
  • 20fbf2a2a0
    ggml : change immintrin.h to intrin.h for compatibility (#1307) master-20fbf2a Ron Jailall 2023-05-04 11:05:59 -04:00
  • f8929309d7
    Download licenses to Henri Vasserman 2023-05-04 18:05:12 +03:00
  • 42b1757522
    Remove testing from matrix Henri Vasserman 2023-05-04 16:37:15 +03:00
  • 530ad68963 print input suffix before generation 44670 2023-05-04 21:23:30 +08:00
  • c08fca9225 adding --in-suffix option 44670 2023-05-04 21:09:04 +08:00
  • db1080876a
    Only escape prompts when used with -e (#1311) master-db10808 DannyDaemonic 2023-05-04 05:08:25 -07:00
  • 795a644962 Avoid hardcoding a space at the beginning of the prompt. Ivan Stepanov 2023-05-04 14:57:55 +03:00
  • 2d5418a69d Wrap exceptions in std::exception to verbose output on exception. Ivan Stepanov 2023-05-04 14:58:52 +03:00
  • f0e44cdeda Remove const char* prompt Danny Daemonic 2023-05-04 04:47:53 -07:00
  • 458aeb10e9 use pause asm insn in busyloop to run the CPU (13600K) 10 °C cooler Sami Farin 2023-05-04 13:51:29 +03:00
  • becce0043e
    Update convert.py Ivan Stepanov 2023-05-04 13:50:43 +03:00
  • e3c2421b9f
    Update convert.py Ivan Stepanov 2023-05-04 13:49:15 +03:00
  • dd8902d3e4
    Update convert.py Ivan Stepanov 2023-05-04 13:46:33 +03:00
  • a3ffcbd98b
    Merge branch 'master' into e-escape DannyDaemonic 2023-05-04 03:12:29 -07:00
  • ccef5e653d Updated README.md example to use -e for Windows prompt Danny Daemonic 2023-05-04 03:06:21 -07:00
  • 1b6f595230 Update main's README.md with new features (#1296) DannyDaemonic 2023-05-04 03:02:59 -07:00
  • 938b7c2e9d fix #1224 reverse prompt and multi line (#1297) Tomas 2023-05-04 17:02:30 +07:00
  • c65a7fbfa9
    Update main's README.md with new features (#1296) DannyDaemonic 2023-05-04 03:02:59 -07:00
  • f647ce040f
    fix #1224 reverse prompt and multi line (#1297) master-f647ce0 Tomas 2023-05-04 17:02:30 +07:00
  • 04c0d480d7
    Move all HIP stuff to ggml-cuda.cu Henri Vasserman 2023-05-04 12:31:16 +03:00
  • d83cfbad0c
    Merge 'origin/master' into hipblas Henri Vasserman 2023-05-04 11:31:16 +03:00
  • 76692c90cd q4_0c: avoid _mm512_loadu_epi64 instruction Håkon H. Hitland 2023-05-04 09:53:55 +02:00
  • b63654c8df load pretrained vocab alex 2023-05-04 09:21:24 +02:00
  • d53f76760d q4_0c: disable prefetching on M1 Håkon H. Hitland 2023-04-27 22:48:46 +02:00
  • 2949725fea q4_0c: prefetch on AVX-512 and ARM Håkon H. Hitland 2023-04-24 18:17:31 +02:00
  • 1b49d26f8a q4_0c: Arm Neon acceleration Håkon H. Hitland 2023-04-21 00:11:49 +02:00
  • ab543dc1a4 q4_0c: AVX512 vec_dot and quantize impl Håkon H. Hitland 2023-04-18 23:07:03 +02:00
  • 4bd781cd25 q4_0c: quantize support Håkon H. Hitland 2023-04-18 00:57:30 +02:00
  • a1e6fb9281 q4_0c continous row layout Håkon H. Hitland 2023-04-17 23:36:29 +02:00
  • 221946777c test-quantize: fix for q8_0 intermediates Håkon H. Hitland 2023-04-16 00:37:16 +02:00
  • c8f7eeb7fd update kobold lite Concedo 2023-05-04 14:43:35 +08:00
  • 981d71b281 Only escape prompts when used with -e Danny Daemonic 2023-05-03 23:23:24 -07:00
  • e01dc631f7 Merge branch 'master' into concedo_experimental Concedo 2023-05-04 14:04:41 +08:00
  • 7c129305f5 derp (+1 squashed commits) Concedo 2023-05-04 12:10:19 +08:00
  • 3f30da38ad llama, main: save state incrementally Evan Jones 2023-05-03 02:09:19 -04:00
  • 866fd3f3cb
    save a token CRD716 2023-05-03 21:19:27 -05:00
  • c47b349281 Support DT_BF16 tensors Ivan Stepanov 2023-05-04 04:09:45 +03:00
  • 932e616cf4
    Code Formatting Tomas 2023-05-04 07:27:52 +07:00
  • 2b7cf9f32b fix too relaxed model glob (breaking multifile) alex 2023-05-04 00:16:12 +02:00
  • aebb5d46ff
    fix typo in ggml.c Ron Jailall 2023-05-03 18:08:54 -04:00
  • 286efed05c
    conditional def of intrin.h Ron Jailall 2023-05-03 18:06:40 -04:00
  • b59c371035 add support for ByteStorage, relax model glob alex 2023-05-03 23:57:08 +02:00
  • ca0a3e78d9
    change immintrin.h to intrin.h for compatibility Ron Jailall 2023-05-03 17:40:33 -04:00
  • 31ff9e2e83
    ci : add cublas to windows release ci_cublas-31ff9e2 ci_cublas Green Sky 2023-05-01 12:41:46 +02:00