root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2025-01-05 16:24:34 +00:00
Code Issues Actions 7 Packages Projects Releases Wiki Activity

Commit Graph

  • 003ba2fb43
         llama : fix hparams shadow (#1367) master-003ba2f Pavol Rusnak 2023-05-08 16:48:21 +0200
  • f9a6364912
         llama : require first token to be BOS (#1303) master-f9a6364 Georgi Gerganov 2023-05-08 17:41:54 +0300
  • 0e94ea65c6
         perplexity : add clarifying comments Georgi Gerganov 2023-05-08 17:41:26 +0300
  • dea9c9359a
         c++ in baby-llama example xaedes 2023-05-08 16:40:31 +0200
  • cdf40a9bea
         readme : update perplexity values after BOS fix Georgi Gerganov 2023-05-08 17:37:49 +0300
  • 10632d820c
         llama: fix shadowing Pavol Rusnak 2023-05-08 15:44:59 +0200
  • 2f2eff6e13 the dark gods have been sated, and redpajama is integrated... but at what cost? Concedo 2023-05-08 20:58:00 +0800
  • afa160d5cc fix typo in default model path Andras Salamon 2023-05-08 13:04:15 +0100
  • 95078cc554
         convert: add ability to convert safetensors files (#1276) ubik2 2023-05-08 04:54:26 -0700
  • d0199b3bc3 fixup! More GPU threads for CUDA kernels JohannesGaessler 2023-05-08 12:56:32 +0200
  • d8c36c91f8
         Check for safetensors files first, and only use PyTorch versions when safetensors aren't available ubik2 2023-05-08 00:56:12 -0700
  • b9904c3093 up ver Concedo 2023-05-08 11:13:16 +0800
  • 1083876a1b Merge branch 'master' into concedo_experimental Concedo 2023-05-08 11:12:42 +0800
  • 4244d97c38
         Merge bcc7f8b6bf into 1f48b0abcf MagisterLuddite 2023-05-08 11:11:30 +0800
  • 89d70886a4 added support for setting custom context size at load time (memory allocation) Concedo 2023-05-08 11:11:25 +0800
  • 1f48b0abcf
         Documented CUDA reproducibility, added warning (#1346) master-1f48b0a Johannes Gäßler 2023-05-08 02:42:01 +0200
  • 1ecbece752
         disable slow tests grad0 and opt to avoid exceeding timeouts xaedes 2023-05-08 02:29:36 +0200
  • f5301061b6
         remove busy loop that was used as sleep for slower sinus wave generation xaedes 2023-05-08 01:12:37 +0200
  • 4997bc5819
         reduce number of test-grad0 iterations xaedes 2023-05-08 00:57:41 +0200
  • 2936dd60a4
         remove trailing whitespace xaedes 2023-05-08 00:04:54 +0200
  • 7c8768f819
         add missing include for strcmp, etc xaedes 2023-05-07 23:42:44 +0200
  • 660836f0ff
         fix call to ggml_set_name xaedes 2023-05-07 23:39:57 +0200
  • 9dd8e405fb
         rename print functions in baby-llama example xaedes 2023-05-07 22:43:23 +0200
  • 47ad186628
         revert disabling of threading for rms_norm and norm xaedes 2023-05-07 21:55:25 +0200
  • 5d9fed7e7f
         remove shape annotations in llama_eval_internal xaedes 2023-05-07 21:45:21 +0200
  • d20ba6f6e6
         update static assert of GGML_OP_COUNT xaedes 2023-05-07 21:42:42 +0200
  • e643fa1619
         smaller default values for baby llama model parameters xaedes 2023-05-07 21:31:55 +0200
  • 10184b0675
         Added TOC and Pygmalion to README AlpinDale 2023-05-07 23:55:00 +0430
  • ee565f34e3
         Merge branch 'master' into train-example xaedes 2023-05-07 21:24:12 +0200
  • 4764842120
         change name of GGML_OP_ADD_AT to GGML_OP_ACC xaedes 2023-05-07 21:14:57 +0200
  • e0de09d77e
         shorten code using a variable xaedes 2023-05-07 19:48:38 +0200
  • 49d6daa11e
         vastly improve training results xaedes 2023-05-07 19:46:05 +0200
  • 93201abdb7
         add trainable lora-only model with all big matrices C split into A,B with A*B=C xaedes 2023-05-07 19:44:51 +0200
  • 8d8de07a4e fixup! More GPU threads for CUDA kernels JohannesGaessler 2023-05-07 18:34:04 +0200
  • 070cbcc1bd
         occupanct function Henri Vasserman 2023-05-07 18:10:56 +0300
  • a3296d50aa
         Merge 'origin/master' into hipblas Henri Vasserman 2023-05-07 18:06:04 +0300
  • 50148408b5 More GPU threads for CUDA kernels JohannesGaessler 2023-05-06 12:17:45 +0200
  • e1295513a4
         CI: add Windows CLBlast and OpenBLAS builds (#1277) master-e129551 Henri Vasserman 2023-05-07 14:20:09 +0300
  • 62beded0e7 Merge branch 'master' into concedo_experimental Concedo 2023-05-07 19:10:01 +0800
  • f12c6f6ec8 ggml : delete unused function, packNibbles_256 katsu560 2023-05-07 20:04:49 +0900
  • ab4f1fa8e2 Output correct error message in ggml_new_tensor_impl when out of scratch space. KerfuffleV2 2023-05-07 03:52:37 -0600
  • 5cc9085353 Works with all characters and control codes + Windows console fixes Danny Daemonic 2023-05-07 02:39:10 -0700
  • d98869197b
         Merge 'origin/master' into cistuff Henri Vasserman 2023-05-07 12:29:29 +0300
  • 963b27450c
         change build name to 'clblast' Henri Vasserman 2023-05-07 12:29:20 +0300
  • 89951748db Add ggml_tensor_required_memory function to calculate how much memory creating a tensor will require. KerfuffleV2 2023-05-07 03:26:58 -0600
  • 0aefa6ab71
         Merge 'origin/master' into hipblas Henri Vasserman 2023-05-07 12:24:41 +0300
  • baeb482a94
         Revert to default copy Henri Vasserman 2023-05-07 12:24:12 +0300
  • 30b2b3d655 There were still struct fields and defines conditionally enabled. KerfuffleV2 2023-05-07 02:25:32 -0600
  • 7f33230a40
         perplexity : add BOS for each chunk Georgi Gerganov 2023-05-07 11:04:52 +0300
  • 1b0fd45465
         ggml : Allow usage of CLBlast alongside Accelerate.framework (#1336) master-1b0fd45 swittk 2023-05-07 10:03:23 +0700
  • 4baa85633a
         Fix build jed/spm-clblast Jed Fox 2023-05-06 21:44:07 -0400
  • 2dc7fc9491
         + ggml-opencl.c Jed Fox 2023-05-06 21:35:00 -0400
  • dc61b7c682
         spm: link with CLBlast Jed Fox 2023-05-06 21:34:13 -0400
  • 0e3d7fd4e9
         Remove default arguments from sampling functions (#1343) Jed Fox 2023-05-06 17:01:47 -0400
  • e91b83b899
         add GGML_ASSERT to catch ggml_rope and back value errors xaedes 2023-05-07 01:47:14 +0200
  • 561fbe0d1b
         replace inplace operations for training with copying operations to allow gradient propagation xaedes 2023-05-07 01:33:42 +0200
  • 956511b248
         fix kv_self gradients for training xaedes 2023-05-07 01:32:46 +0200
  • 47561de7d8
         add ggml_set(ctx, a, b) to set b in view of a and return modified a xaedes 2023-05-07 01:30:34 +0200
  • 48bcc4dcf9
         fix backward pass for add_at and change arguments to have same order as in view xaedes 2023-05-07 01:27:11 +0200
  • 226521a4f1
         optimize loss over multiple samples xaedes 2023-05-07 01:23:51 +0200
  • 7a5dec24f8
         add square_error_loss and cross_entropy_loss functions xaedes 2023-05-07 01:21:26 +0200
  • 73fd66e9e5
         fix training get_example_targets xaedes 2023-05-07 01:18:17 +0200
  • 7523107877 Keep API and context fields the same whether or not GGML_RECOVERABLE_ERRORS is defined. KerfuffleV2 2023-05-06 15:48:57 -0600
  • 1a6987ad92 First rough draft of recoverable errors feature. KerfuffleV2 2023-05-04 14:42:30 -0600
  • 3924088512
         Remove default arguments from sampling functions (#1343) master-3924088 Jed Fox 2023-05-06 17:01:47 -0400
  • 4b18cdf632 Documented CUDA reproducibility, added warning JohannesGaessler 2023-05-06 20:53:47 +0200
  • 289073a532
         Merge 'origin/master' into hipblas Henri Vasserman 2023-05-06 19:59:41 +0300
  • 80223d98fd
         add test for ggml_sum_rows gradients xaedes 2023-05-06 18:01:32 +0200
  • e6186d98a5
         implement ggml_repeat support for rank > 2 tensors xaedes 2023-05-06 18:01:17 +0200
  • fb9e1dde5c
         Add missing parameters to examples & tests Jed Fox 2023-05-06 11:46:44 -0400
  • 09236f40b4
         llama license text Henri Vasserman 2023-05-06 18:41:56 +0300
  • 7a15a8370c
         implement backward pass for ggml_sum_rows, necessary for cross entropy loss xaedes 2023-05-06 17:37:51 +0200
  • 5724628d31
         add test for ggml_log gradients xaedes 2023-05-06 17:36:21 +0200
  • 65d9f7349d
         add ggml_log operation necessary for cross entropy loss xaedes 2023-05-06 17:35:13 +0200
  • 8cf04fec9d
         fix soft_max backward pass for input->ne[1] != 1 xaedes 2023-05-06 17:30:38 +0200
  • b4c273f7a3
         add ggml_reshape_1d, ggml_reshape_4d and ggml_view_4d xaedes 2023-05-06 17:29:41 +0200
  • 2986951203
         fix Henri Vasserman 2023-05-06 18:02:05 +0300
  • 0dfa17db1e
         fix Henri Vasserman 2023-05-06 17:59:44 +0300
  • 87d8ac9286
         fix Henri Vasserman 2023-05-06 17:59:26 +0300
  • 5cb13c2fcc
         add version numbers Henri Vasserman 2023-05-06 17:57:02 +0300
  • 71fac5bbcb
         Merge 'origin/master' into cistuff Henri Vasserman 2023-05-06 16:57:21 +0300
  • 46879f35a6
         Remove default arguments from sampling functions Jed Fox 2023-05-06 09:25:43 -0400
  • 5d4640511b
         Add *.bin to gitignore Jed Fox 2023-05-06 09:25:02 -0400
  • f1d51d144b
         train on multiple examples, generate & print tokens with trained model afterwards xaedes 2023-05-06 14:13:55 +0200
  • 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 +0200
  • 23f148516d use _mm_pause() in busyloop Sami Farin 2023-05-06 07:48:45 +0300
  • ff93b394da fixed a typo Concedo 2023-05-06 12:37:34 +0800
  • a48dddab86 slightly bump the RAM up to support chinese alpaca Concedo 2023-05-06 11:48:22 +0800
  • 1fa3128d2a
         Allow usage of CLBlast alongside Accelerate.framework swittk 2023-05-06 08:24:48 +0700
  • fe72648fac feat: add alibi Zach Nussbaum 2023-05-05 20:21:04 -0400
  • 173d0e6419
         makefile: automatic Arch Linux detection (#1332) master-173d0e6 DaniAndTheWeb 2023-05-05 23:57:14 +0200
  • 1107194e6b
         Merge 'origin/master' into hipblas Henri Vasserman 2023-05-06 00:38:20 +0300
  • a44c384c56
         Shrink Arch Linux check DaniAndTheWeb 2023-05-05 23:23:47 +0200
  • ab5e61d710
         Fix possible error DaniAndTheWeb 2023-05-05 23:10:28 +0200
  • a3b85b28da
         ci : add cublas to windows release (#1271) master-a3b85b2 Erik Scholz 2023-05-05 22:56:09 +0200
  • 471624b464
         Rename variable DaniAndTheWeb 2023-05-05 22:52:49 +0200
  • c0a86cc556
         Optimize Arch Linux check DaniAndTheWeb 2023-05-05 22:51:56 +0200
  • e0ed30dffe Dockerfiles: use LLAMA_AVX2_ONLY. Bryan Silverthorn 2023-05-05 10:49:03 -0700
  • 2e3b2e7e09 Makefile: support LLAMA_AVX2_ONLY. Bryan Silverthorn 2023-05-05 10:48:13 -0700
  • 8a964e76c8 integrated mirostat as a launch parameter, works on all models Concedo 2023-05-06 00:47:17 +0800