Commit Graph

  • 9f8bbd3fee readme : update comment about source code Georgi Gerganov 2023-11-12 17:47:37 +02:00
  • 3172006a24 ggml : fix some compile warnings Georgi Gerganov 2023-11-12 16:36:20 +02:00
  • 684bc8bd70 readme : update GPU / CUDA Georgi Gerganov 2023-11-12 15:40:37 +02:00
  • b0502836b8 whisper : add full CUDA and Metal offloading (#1472) Georgi Gerganov 2023-11-12 15:31:08 +02:00
  • 5031f54717 whisper : try to fix the parallel whisper_state functionality (#1479) ggml-backend-no-sched Georgi Gerganov 2023-11-12 14:52:38 +02:00
  • a2f3b82db3 whisper : free backend instances in whisper_state parallel-states Georgi Gerganov 2023-11-12 14:31:51 +02:00
  • 76c8b5235b whisper : fix multi-state Metal Georgi Gerganov 2023-11-12 14:24:45 +02:00
  • d029784fb0 whisper : try to fix the parallel whisper_state functionality Georgi Gerganov 2023-11-11 18:37:14 +02:00
  • 40c66036b6 whisper : fix UB with measure buffers Georgi Gerganov 2023-11-11 18:35:23 +02:00
  • fc8565d0e2 whisper : fixes Georgi Gerganov 2023-11-11 17:39:30 +02:00
  • b618229340 whisper : factor out graph compute in common function Georgi Gerganov 2023-11-11 17:06:21 +02:00
  • b27726da93 whisper : add note that ggml_mul_mat_pad does not work with CUDA Georgi Gerganov 2023-11-11 13:04:58 +02:00
  • 0867e696a7 whisper : avoid whisper_model_data wrapper Georgi Gerganov 2023-11-11 11:46:54 +02:00
  • 66bb2e9401 ggml : im2col opts Georgi Gerganov 2023-11-11 10:41:00 +02:00
  • 3bfc43e3e3 quantize-all : fix Georgi Gerganov 2023-11-10 23:33:40 +02:00
  • f53e1388f5 whisper : clean-up Georgi Gerganov 2023-11-10 22:31:44 +02:00
  • 933c5bef97 whisper : support ggml_conv with CUDA and Metal (#1473) Georgi Gerganov 2023-11-10 22:26:50 +02:00
  • 7a91a3ba60 bench-all : add q4 models ggml-conv Georgi Gerganov 2023-11-10 22:23:18 +02:00
  • 997f7cbf13 metal : add im2col support + mul mat-vec f16 x f16 Georgi Gerganov 2023-11-10 21:32:21 +02:00
  • 9c1ddc77a7 cuda : fix im2col kernel Georgi Gerganov 2023-11-10 19:39:24 +02:00
  • 000b952c2d whisper : remove ggml_repeat for conv bias + single backend Georgi Gerganov 2023-11-10 15:54:43 +02:00
  • 81506268ba ggml : add CUDA support for ggml_conv Georgi Gerganov 2023-11-10 15:10:27 +02:00
  • c99e290a7f talk : fix compile warning Georgi Gerganov 2023-11-10 13:54:02 +02:00
  • 728e1785f0 Merge branch 'master' into ggml-backend-no-sched Georgi Gerganov 2023-11-10 13:51:31 +02:00
  • ec7a6f04f9 whisper : return with error from whisper_encode_internal and whisper_decode_internal when abort callback is true (#1456) Ben Nortier 2023-11-10 13:51:16 +02:00
  • d6dad64fbf make : clean-up Georgi Gerganov 2023-11-10 13:45:07 +02:00
  • a54d8c9dec whisper : fix CoreML Georgi Gerganov 2023-11-10 13:24:06 +02:00
  • 0ab5025316 Merge branch 'master' into ggml-backend-no-sched Georgi Gerganov 2023-11-10 13:21:47 +02:00
  • 3f5c1b7ee0 whisper : print when CUDA is enabled Georgi Gerganov 2023-11-10 13:17:02 +02:00
  • 12030358ee whisper : free backends + fix compile warning Georgi Gerganov 2023-11-10 12:45:26 +02:00
  • dcf9511dbb whisper : fix beam-search with CUDA Georgi Gerganov 2023-11-10 12:41:11 +02:00
  • 3dfbe64911 whisper : fix tensor allocation during load Georgi Gerganov 2023-11-10 11:51:55 +02:00
  • 7e01486b61 whisper : fix logit reading Georgi Gerganov 2023-11-10 11:02:29 +02:00
  • 659757329d whisper : migrate to ggml-backend Georgi Gerganov 2023-11-10 10:54:06 +02:00
  • 37947203e6 talk-llama : add language auto detect (#1467) Jakub Ráček 2023-11-09 18:21:44 +01:00
  • bf4110dbcf whisper : wip sched (not working yet) ggml-backend Georgi Gerganov 2023-11-09 19:07:54 +02:00
  • 005b8ccbf0 whisper : migrating to ggml-backend (wip) Georgi Gerganov 2023-11-08 14:53:24 +02:00
  • 953419c69a openvino : update convert-whisper-to-openvino.py to support v3 (#1459) bobqianic 2023-11-09 10:42:39 +00:00
  • 0de8582f65 coreml : use the correct n_mel value (#1458) Xiao-Yong Jin 2023-11-08 14:01:41 -06:00
  • baeb733691 whisper : reset mel time when resetting timings (#1452) Ben Nortier 2023-11-08 15:52:23 +02:00
  • d03c60dd7f ios : add support for Swift Package Manager (#1370) Sindre Sorhus 2023-11-08 04:53:31 +07:00
  • 6a5d195109 release : v1.4.3 v1.4.3 Georgi Gerganov 2023-11-07 16:15:48 +02:00
  • 0cbef75422 ggml : fix MIN / MAX macro re-definition Georgi Gerganov 2023-11-07 16:08:46 +02:00
  • 2cdfc4e025 whisper : add support for large v3 (#1444) Georgi Gerganov 2023-11-07 15:30:18 +02:00
  • 973111088b android : decouple example into a library and app module (#1445) Tobrun 2023-11-07 13:27:33 +01:00
  • 40be74271f models : update readme large-v3 Georgi Gerganov 2023-11-07 13:53:01 +02:00
  • a0c0d08d0f bench : fix n_mels Georgi Gerganov 2023-11-07 13:45:56 +02:00
  • 8fb0a1cd1c bench : fix build + fix go bindings Georgi Gerganov 2023-11-07 13:20:02 +02:00
  • 185d3fd6d9 whisper : add support for large v3 Georgi Gerganov 2023-11-07 11:58:39 +02:00
  • 11b503055e whisper : reset ctx->t_start_us when calling whisper_reset_timings() (#1434) Ben Nortier 2023-11-07 11:04:32 +02:00
  • b629d2d4fe cmake : fix talk-llama build Georgi Gerganov 2023-11-07 11:03:21 +02:00
  • 3bd7d48f51 metal : fix asserts for setThreadgroupMemoryLength (close #1435) Georgi Gerganov 2023-11-07 11:02:16 +02:00
  • 435a6b74e3 ci : fix variable names in GitHub actions config (#1440) iamthad 2023-11-07 02:53:24 -06:00
  • 75dc800d21 talk-llama : fix n_gpu_layers usage again (#1442) Jhen-Jie Hong 2023-11-07 16:51:27 +08:00
  • 0c91aef2d8 whisper : add missing about callback initializers Georgi Gerganov 2023-11-07 10:49:51 +02:00
  • 3989b29a9b examples : fix n_gpu_layers usage in talk-llama (#1441) Jhen-Jie Hong 2023-11-07 09:36:23 +08:00
  • 0463028bc2 whisper : add context param to disable gpu (#1293) Jhen-Jie Hong 2023-11-06 17:04:24 +08:00
  • aaa3b5e5f6 ggml : try to fix the abort mechanism try-fix-abort Georgi Gerganov 2023-11-05 20:02:24 +02:00
  • 39cfad0dee whisper : add support for new distilled Whisper models (#1424) Georgi Gerganov 2023-11-05 19:43:45 +02:00
  • 673c55c683 whisper : print log when using distilled models distil-support Georgi Gerganov 2023-11-05 19:43:04 +02:00
  • 6d4d0b5b4b cuda : fix HIPBLAS build Georgi Gerganov 2023-11-05 19:41:15 +02:00
  • f96e1c5b78 sync : ggml (backend v2, k-quants, CUDA opts, Metal opts, etc.) (#1422) Georgi Gerganov 2023-11-03 21:35:05 +02:00
  • b8c93c5f3b whisper : add support for new distilled Whisper models Georgi Gerganov 2023-11-03 13:24:05 +02:00
  • 8a2bee6717 models : use absolute paths for the converted model (#1356) bobqianic 2023-11-03 08:44:27 +00:00
  • d445098c8f talk-llama : move up-to-date demo to top (#1417) Asad Memon 2023-11-02 09:50:13 -07:00
  • 74de25158e talk-llama : add an up-to-date demo video Georgi Gerganov 2023-11-02 15:28:48 +02:00
  • bce49a260e examples : Implement JSON output for Token-Level data in main (#1358) Aarni Koskela 2023-10-31 21:54:52 +02:00
  • 45c87b5481 models : Faster download for models on windows using BitTransfer (#1404) WhiteOlivierus 2023-10-30 20:18:12 +01:00
  • dfe4bc6e59 README : Update README in stream to clarify where to compile from (Issue #1400) ai-at-home 2023-10-29 12:11:13 -05:00
  • 54c978c3a3 binding : Expose the audio_ctx param through the Go binding (#1368) Johan 2023-10-15 05:35:06 -07:00
  • 9a7074d4aa README : fix typo (#1362) jorismertz 2023-10-13 17:53:23 +02:00
  • a0040f5d12 docker : Add dockerfile for cublas (#1286) joecryptotoo 2023-10-11 03:00:17 -07:00
  • 940cdb1396 whisper : abort callback improvements (#1345) mkiol 2023-10-08 16:22:24 +02:00
  • 1b775cdd68 cmake : Abort the build if a requested feature could not be configured (#1350) Marcin Mielniczuk 2023-10-07 21:01:18 +02:00
  • 80bf931668 cmake : Prefer pkg-config while looking for BLAS (#1349) Marcin Mielniczuk 2023-10-07 16:02:07 +02:00
  • 91c0b23384 models : add conversion scripts from HuggingFace models to CoreML (#1304) Xiang (Kevin) Li 2023-10-04 05:00:25 -04:00
  • 2f668c330e whisper : add abort callback (#1335) mkiol 2023-10-04 10:57:55 +02:00
  • 08fa34882f examples : move wav_writer from stream.cpp to common.h (#1317) bobqianic 2023-10-03 20:56:11 +01:00
  • 4037705531 whisper : add missing speaker turn API function for whisper_state (#1330) Didzis Gosko 2023-10-03 22:55:48 +03:00
  • c76c11e59c examples: Update the README for Talk - fixing the gpt2 URL (#1334) brunofaustino 2023-09-30 17:21:32 -03:00
  • 9edbd0a204 extra: Add benchmark script implemented in Python (#1298) Neil Chudleigh 2023-09-25 08:45:15 -07:00
  • 707507ff6d Examples: Add save audio to file option in stream.cpp (#1310) litong 2023-09-22 05:43:21 -10:00
  • 7e1592d2cd readme: Fix spelling error (#1290) JJ 2023-09-21 00:55:33 -07:00
  • 903c9579b8 examples: Update README.md of main.cpp (#1306) Artyom Mezin 2023-09-18 17:14:36 +03:00
  • b440ef8c96 binding : fix ruby build by adding missing ggml-alloc (#1305) Jhen-Jie Hong 2023-09-18 21:15:45 +08:00
  • 700f63a806 bench: fix missing include <cstring> (#1303) Evgeny Kuznetsov 2023-09-18 10:51:10 +03:00
  • 951a119926 whisper : increase tokenizer buffer (close #1259) Georgi Gerganov 2023-09-15 21:11:43 +03:00
  • 1ca4041b86 talk-llama : update to latest llama.cpp Georgi Gerganov 2023-09-15 20:06:31 +03:00
  • 80c1512fd5 sync : ggml (const correctness) Georgi Gerganov 2023-09-15 14:49:56 +03:00
  • 0ac9cefd03 metal : restore matrix x vector f16_f32 kerenls for now Georgi Gerganov 2023-09-15 14:40:41 +03:00
  • b8432f28f4 metal : add F32 support + update bench output Georgi Gerganov 2023-09-15 13:56:08 +03:00
  • 93935980f8 whisper : Metal and ggml-alloc support (#1270) Georgi Gerganov 2023-09-15 12:18:18 +03:00
  • 3ac0558009 ios : update SPM package metal-and-alloc Georgi Gerganov 2023-09-15 12:13:33 +03:00
  • a1664574fe bench : variable n_past Georgi Gerganov 2023-09-14 22:41:41 +03:00
  • bfcb2a2ab9 metal : remove the "concurrent" flag Georgi Gerganov 2023-09-14 18:04:42 +03:00
  • 0d5e4cdc36 whisper : clean-up ggml_mul_mat_pad Georgi Gerganov 2023-09-14 17:28:13 +03:00
  • 2b4160af29 whisper : add description of ggml_mul_mat_pad Georgi Gerganov 2023-09-14 15:36:36 +03:00
  • f36554382a whisper : add comment for disabling mul-mat padding Georgi Gerganov 2023-09-14 15:25:19 +03:00
  • c46167f8c5 bench : fix uninitialized vars Georgi Gerganov 2023-09-14 15:19:27 +03:00
  • af947cb72e whisper : add ggml_mul_mat_pad Georgi Gerganov 2023-09-14 15:16:22 +03:00