f131bf1908b8c5dd1f49d7ae7f616506fc471666	33d96dae28e17208ef61a71dba40cda3c04f135a	2026-05-27T09:38:51+08:00	wangbomeng		fix patch of qwen2.5vl,n_patch of minicpm = 64,fix n_ctx_slot,comment out M-ROPE
2716130010c58f1cdd9d94f0e2f8e047bc92dc66	b35bda124ccd4ed581dd6088d3ffa4931c2809b6	2026-05-18T08:40:54+08:00	Yunzhe Jia		fix n_ctx_slot error by adding runtime capacity loading for cal-llm
ecf01a17651cd7b1e8b85fd6075e83831463ee23	99d2078e89305035d87d966cee7987c7d50b3925	2026-04-09T09:47:55+08:00	Yunzhe Jia		fix n_parallel problem
1f5accb8d0056e6099cd5b772b1cb787dd590a13	2759ccdb4adc8568add4316780d5e675519b0775	2025-11-04T05:04:59Z	Noah		Fix garbled output with REPACK at high thread counts (#16956)
bcfa87622ae46be6345a8e3dfdbdc5ba5414042b	a2054e3a8ff0da3978a4acc18c349ff58554d336	2025-11-03T00:41:08+01:00	Sascha Rogmann		feat(webui): improve LaTeX rendering with currency detection (#16508)
cd5e3b57541ecc52421130742f4d89acbcf77cd4	87c9efc3b297b8a498716b1db3d061842e6fc85b	2025-11-02T18:14:04+02:00	Georgi Gerganov		server : support unified cache across slots (#16736)
2f68ce7cfd20e9e7098514bf730e5389b7bba908	e4a71599e5846110159955dec0008eb4aa24222b	2025-11-01T19:49:51+01:00	Pascal		webui: auto-refresh /props on inference start to resync model metadata (#16784)
d3dc9dd898be805c23a408cc36daed5b3bf29221	bea04522ff1a0d8559ccfd353aa018dcfbb608cc	2025-11-01T06:13:26+01:00	Oliver Simons		CUDA: Remove unneded bias/gate dims in fused mmvq (#16858)
0de0a01576772032008a689afc4d7c80685074c4	e58d585604bbbbbc24f8149effe444621b5191c6	2025-10-31T21:20:47+01:00	Piotr Wilkin (ilintar)		model : Minimax M2 (#16831)
b52edd25586fabb70f0c21b274473b307cf14499	517b7170e1a4d733583c4b07c5b7a49acc05911c	2025-10-30T18:42:57+02:00	Georgi Gerganov		server : remove n_past (#16818)
8b11deea4663f29d3e042ce1056ba643264cd5f1	b9ce94017729465895402cbcfffb51fa926c15e3	2025-10-30T04:34:15+01:00	Oliver Simons		Hide latency of bias and gate-loading (#16847)
b9ce94017729465895402cbcfffb51fa926c15e3	3464bdac37027c5e9661621fc75ffcef3c19c6ef	2025-10-29T15:13:10-05:00	Jeff Bolz		vulkan: Fuse rope+set_rows (#16769)
bcf5bda6f5df559565d11d7c8e8295c1159a85ec	3eb2be1ca5f37480aeb16102970d9e65f43347fe	2025-10-29T14:39:03+01:00	Ruben Ortlam		Vulkan MMQ Integer Dot Refactor and K-Quant support (#16536)
3eb2be1ca5f37480aeb16102970d9e65f43347fe	e41bcce8f0b53032a1fed275cd253e931c041cf6	2025-10-29T06:29:12-07:00	Max Krasnyansky		Hexagon Op queue & dispatch optimizations (#16820)
338074c383c81366320d176d83b94b0a567ee0c2	851553ea6b24cb39fd5fd188b437d777cb411de8	2025-10-29T08:14:39+02:00	YaelLogic		sycl: add RMS_NORM_BACK operation support (#16808)
85a7d8677bf2200981e52f744a21d5267964ffcf	a8ca18b4b815a2abdbecb958ee5f4c542d69aac7	2025-10-28T20:19:44+02:00	Georgi Gerganov		memory : remove KV cache size padding (#16812)
1c1409e13169d0ced4b1b4d39fb5b268b7525091	7a0e900e3615fa46c074a7fdf900b47d3c0a1c7e	2025-10-28T03:51:41-07:00	Sam Malayek		embedding: add raw option for --embd-output-format (#16541)
3479efd112b3910d2d008ad08fe4cb4895605903	463bbf20bfbe0da10ac58984d4d62bed5a49362a	2025-10-28T10:54:53+08:00	Chenguang Li		CANN: Improve device ID handling and aclnnArange checks (#16752)
ad8d36beffd791db10c94eb9e964afb891e3ca55	c053e18a66dd95dc340aa61317877c2a41d4e3cf	2025-10-28T03:50:33+02:00	tamarPal		sycl: add SSM_CONV operation support (#16800)
5a4ff43e7dd049e35942bc3d12361dab2f155544	10640e31aab0819f31c1e1f2d008b019ee737232	2025-10-27T13:51:28-07:00	Diego Devesa		llama : disable pipeline parallelism if compute buffer allocation fails (#16748)
55945d2ef51b93821d4b6f4a9b994393344a90db	0bcb40b48c6fc6f17ba9672625e526ab2574344b	2025-10-25T03:39:37+08:00	leejet		ggml: fix CUDA grid launch condition for large block_nums.y in binbcast (#16742)
fe6a9882acf5c02f96624ed8f80144100d7006cb	061f0eff02dc9a82f7bd850db3bd70b8a0b5e87a	2025-10-23T17:07:31+05:30	Prajwal B Mehendarkar		Manually link -lbsd to resolve flock symbol on AIX (#16610)
9de9672adb0f4ca4e39483ac3ffed52b3f70a55d	63d2fc46e17a06be5b4b5823a5ada088317f1f0a	2025-10-22T20:05:15-05:00	Matthew Michel		sycl: use async memory allocation to fix crashes during graph recording (#16644)
d8eaa26e4d9228df3aa46a930db60c8eaab67c1b	9285325ce0174631c3cd6121d56084adc4ef2d8f	2025-10-22T12:01:22+02:00	Acly		tests : fix test-thread-safety when compiling with multiple backends (#16699)
6de8ed75196c7cd98c1f34bbf3a7452451ba8ac2	84bf3c677857279037adf67cdcfd89eaa4ca9281	2025-10-21T01:21:12+03:00	YehuditE		sycl : add PAD_REFLECT_D1 operator support (#16145)
13f2cfad4170c096c51a02c24a6a158cb47f1480	06332e28672356b964d6dfc2ba4657e20581cd43	2025-10-20T12:41:13+02:00	Aleksander Grygier		Enable per-conversation loading states to allow having parallel conversations (#16327)
06332e28672356b964d6dfc2ba4657e20581cd43	72d53e6c4decee8b339e49aed8cc0e234b9639dc	2025-10-20T16:27:09+08:00	takuya kodama		llama-batch: fix build fails with `-Werror=missing-braces` (#16614)
342c728d031d50673feded797520a44127d73379	ababae7e1ec3e9cfdab0322ee55ea3389e82a4d5	2025-10-17T18:01:23+08:00	muggle-stack		ggml : fix SpaceMit IME array out-of-bounds in task assignment (#16629)
9ad4f1931ee0f3b41d9355245ef744786aaae0aa	79967ec596c0dacfd2251b085a57e79df292b1cc	2025-10-17T02:33:58-04:00	Ilia Ilmer		metal : add `CONV_TRANSPOSE_2D` (#16542)
683fa6ba4ed3a23b939d3e11e6dc860bd47a0ccf	b22572e97dc51757d3ebe917a5a283385010ec68	2025-10-16T16:28:41+02:00	Pascal		fix: added a normalization step for MathJax-style \[\] and \(\) delimiters (#16599)
6f5d924637a15abedb111cbbffd7da5f31c81855	adc9b60f190c1016a09f439862fa1cbb302262ac	2025-10-16T01:11:33-04:00	takasurazeem		common : Update the docs on -t --threads (#16236)
adc9b60f190c1016a09f439862fa1cbb302262ac	ee50ee1eadff58777ae746827b04de7ba0befc55	2025-10-16T13:10:32+08:00	takuya kodama		ggml-cpu: replace putenv with setenv for const-correctness (#16573)
7ea15bb64c81e3813eb0babf9a57e1bc5697f569	9c7185dd28416cf67f5e3b268381f311b5e3da56	2025-10-14T07:51:36-05:00	Jeff Bolz		vulkan: Improve build time for MSVC (#16545)
1fb9504eb744969a990bfe4cfcf1d3d7a479541c	3f750f8d760ab5a61491e6a9409072dfeee4b4d7	2025-10-13T10:55:32+02:00	Pascal		fix: add remark plugin to render raw HTML as literal text (#16505)
f9bc66c3ebcfddb5f09e4b21253623caeb8e414a	a31cf36ad946a13b3a646bf0dadf2a481e89f944	2025-10-13T08:52:22+08:00	hipudding		CANN: Update several operators to support FP16 data format (#16251)
d00cbea63c671cd85a57adaa50abf60b3b87d86f	8328fd4bae76fc027f8ca0e9a05acd3788dabe3f	2025-10-09T18:54:51+03:00	Georgi Gerganov		server : host-memory prompt caching (#16391)
e08db4259521de493b7aeb49dadf29ebd1ee966a	12bbc3fa50b6df03318a4451c9a2210200a0b28d	2025-10-09T08:39:18+02:00	Saba Fallah		model: EmbeddingGemma Adding Support for SentenceTransformers Dense Modules (#16367)
12bbc3fa50b6df03318a4451c9a2210200a0b28d	9d0882840e6c3fb62965d03af0e22880ea90e012	2025-10-08T22:18:41+02:00	Pascal		refactor: centralize CoT parsing in backend for streaming mode (#16394)
b2c08c9ec4cd59ed88c1ebdd94f109af5ccc978e	7fdd16b432d247121fe5fe7b21f8805f85266c85	2025-10-08T10:57:53+03:00	Georgi Gerganov		metal : mark FA blocks (#16372)
74b8fc17f92ada295a648e3c5eb28f46bca7d892	aeaf8a36f06b5810f5ae4bbefe26edb33925cf5e	2025-10-07T13:48:56-07:00	Reese Levine		ggml webgpu: profiling, CI updates, reworking of command submission (#16452)
1d6092fc72f4d10f4486ac95edfd414bc08b62b8	8ae32dc9ecd7aeebf3a5b43557e0552a0a04cd4f	2025-10-07T08:22:35+03:00	Georgi Gerganov		tests : add -INF blocks to the KQ mask in the FA tests (#16380)
ca71fb9b368e3db96e028f80c4c9df6b6b370edd	35266573b968e1c947b367782fb4b3eddbb4f3c0	2025-10-05T06:57:47-06:00	Gabe Goodhart		model : Granite docling + Idefics3 preprocessing (SmolVLM) (#16206)
35266573b968e1c947b367782fb4b3eddbb4f3c0	86df2c9ae4f2f1ee63d2558a9dc797b98524639b	2025-10-04T20:59:31-07:00	Reese Levine		ggml webgpu: actually add softmax, fix rms_norm offset (#16400)
86df2c9ae4f2f1ee63d2558a9dc797b98524639b	f39283960b58a92ecc0c72567711318b20e22b55	2025-10-04T20:04:27Z	Eve		vulkan: use a more appropriate amount of threads when generating shaders (#16418)
0e1f83855609d73beaf05d818640b6cfd39d287b	ad126479c25cf983a0f994a08ba0911cf49ed62b	2025-10-03T04:52:46-05:00	Jeff Bolz		vulkan: Fix FA coopmat1 invalid array indexing (#16365)
d64c8104f090b27b1f99e8da5995ffcfa6b726e2	ef07a4090672a3438d7f64f197795d7dc1c18957	2025-10-02T20:10:12+02:00	Sigbjørn Skjæret		test-barrier : do not use more threads than physically available (#16389)
2a9b63383a448ec18c754dfdc6e95cb853940a52	1104ca1a1c926b832274694a2773a17066982ad0	2025-10-01T15:54:42+02:00	Aleksander Grygier		Improve code block color theming (#16325)
f1eb1cb1eba042b2d583234e80f65e2e225d8995	de41f2b7bffab7582944b213ce12b605f8cf6e0f	2025-09-30T09:07:20+02:00	Charles Xu		kleidiai : fix work size and threads sync for fp16 (#16246)
5f7e166cbf7b9ca928c7fad990098ef32358ac75	d72f5f7ba260b546190338b0b76f2f152581424f	2025-09-29T18:49:47+02:00	Pascal		Fix thinking blocks with quotes + add handling `[THINK]...[/THINK]` blocks (#16326)
c0bfc57af421f8fd63c946c13b7666aed82560e2	75a3a6c2cd0002ba40e2dcc92007bc9fdbc69f1a	2025-09-28T00:49:32+08:00	Aman Gupta		CUDA: mul_mat_id for mmf for bs <= 64 for f16 and bs <= 32 for f32 (#16277)
624207e676ab5eb3ce7af631902bb45fb73a8359	807e8c6d310952f2f5656afc63dab9d7083dcb5c	2025-09-27T02:03:33+08:00	Aaron Teo		devops: add s390x & ppc64le CI (#15925)
9b26511857ac09ae69ab485168fe2d3ee5fb1d6e	00217cd41388328c93e9e9644921c4319bb03bcc	2025-09-26T18:27:25+08:00	Aaron Teo		ggml-cpu: implement MXFP4 SIMD for s390x (#16193)
a86a580a6692edd1da076d5422f944c344b4fe5e	0f7c69689f4e681948a6005adea9bee9c08ff903	2025-09-26T08:56:38+08:00	R0CKSTAR		musa: upgrade musa sdk to 4.3.0 (#16240)
0f7c69689f4e681948a6005adea9bee9c08ff903	835b2b915c52bcabcd688d025eacff9a07b65f52	2025-09-26T08:56:10+08:00	R0CKSTAR		musa: fix build warnings (#15611)
f2a789e33490deb483a2694b066b37e45524bb79	3a599719673c850647e3bb911ed6d91109bb91d2	2025-09-24T16:17:49+02:00	Acly		ggml : split graph allocations according to backend max buffer size (#15815)
3a599719673c850647e3bb911ed6d91109bb91d2	63b54c81a620981be020184ab99e63a8e50e47cb	2025-09-24T13:42:26+02:00	Tarek Dakhran		model : add label for LiquidAI LFM2-2.6B model (#16204)
1eeb523c3e0c7ffbd59469f5463dcbdecba3535e	5bb4a3edec297e74b0f7bd4ed5d0fdd12e28d858	2025-09-21T08:31:55+02:00	Giuseppe Scrivano		vulkan: optimize UMA buffer operations and fix driver hangs (#16059)
69ffd891631befa9e6b485fd646a16dab4f2c007	246c0d9c795fb25da8268625d597580155a3672f	2025-09-18T23:07:26+02:00	Adrien Gallouët		ggml-amx : fix ggml_amx_init() on generic Linux (#16049)
d304f459d8619399eb232b1e3689379306f439d3	0320ac5264279d74f8ee91bafa6c90e9ab9bbb91	2025-09-17T13:09:40-07:00	Reese Levine		GGML WebGPU: Support for ADD, MUL, RMS_NORM, GET_ROWS operators (#16018)
cd08fc3ecc0264b4414b68af3874a6c689ed60c1	cb5bb6cc05119c24e7711ca2956cd0e6d409d396	2025-09-17T01:08:02-07:00	David Ribeiro Alves		common : Fix corrupted memory error on json grammar initialization (#16038)
d5fabe3682de515fd09d6c981f7a0d1b75614455	8ff206097c2bf3ca1c7aa95f9d6db779fc7bdd68	2025-09-17T14:33:08+08:00	Chenguang Li		CANN: Optimize ggml_cann_set_device (#15935)
f161463a54d9f93d41246286aa4a9569a91d804d	84d7b2fca11d1be118ce776f6d72a486c4883b74	2025-09-13T13:54:28+03:00	Georgi Gerganov		metal : allow ops to run concurrently (#15929)
704d90c987cdf00751567b2088c4e54742aa2d3f	360d6533db39e11577afe9b0aece20c6b5ddaf1f	2025-09-12T09:15:12+08:00	Neo Zhang Jianyu		Revert "sycl: add usage of enqueue_functions extension (#14244)"  (#15910)
0f0a3c2851134d49955f3c85afbb0b1bb47c3e07	33daece86b65607451d0d4378d2d04ba6a20ad55	2025-09-10T17:52:35+03:00	Georgi Gerganov		metal : make the backend async (#15906)
a972faebed5fdc4a3d2a844d92d476058c02e02d	550cf726e133fd0a069d991287fd3a2a3e3e1cbd	2025-09-09T14:38:02+08:00	Aman Gupta		CUDA: Add mul_mat_id support for the mmf kernel (#15767)
e68aa10d8f3d26fdad5b912540362d79de5460e3	0a16bf52e6874369cb4fd2bdc4863ef984b688e7	2025-09-08T13:10:07-05:00	Jeff Bolz		vulkan: sort graph to allow more parallel execution (#15850)
b0d52998b962bd2681c34bf52af993af79f178b8	f28d4f4ac963f182ea9d0fe9e269f3f5f3782aaf	2025-09-08T13:56:51+03:00	Georgi Gerganov		cuda : fix supports_op condition for get_rows when number of blocks is too large (#15868)
fd621880f3fd908424e675a41715a2dc760247a2	4281c7b315f8a904549fe527039b976e08098d1a	2025-09-05T17:32:39-06:00	Gabe Goodhart		aLoRA Support (#15327)
d1e2adba65208d6de3d7f6f23b00c562ff5bb777	c1c354e44c06d259679bb5bb7a8fa9f0b28480e4	2025-09-04T15:40:44+02:00	Daniel Bevenius		llama : set n_outputs to 1 to avoid 0 outputs mean-pooling (#15791)
c1c354e44c06d259679bb5bb7a8fa9f0b28480e4	a68d9144262f1d0ef4f6ba7ad4a7e73e977ba78c	2025-09-04T20:20:14+08:00	Chenguang Li		CANN: Refactor ND to NZ workspace to be per-device (#15763)
661ae31c9c68201577e70278285b349a5a662caf	407c23786dd0d3a503e9429eead96e611d3950c9	2025-09-03T19:59:16+02:00	Oliver Simons		CUDA: Optimize `rms_norm_f32` kernel and its fused variants, giving 1-6% perf E2E (#15715)
02c1813517412f3e00aa6ca7c0273fea64edb492	77dee9de97be75b7143a213bc48893e0c0b29af7	2025-09-01T16:19:07+02:00	Ruben Ortlam		Vulkan: Add Integer Dot Product mul_mat_vec shader for legacy quants (#14903)
0d161f021aa33ec0e90cce96f5d1a88925557327	4efd5a83163ff383285b3a4c2106feabf5c69557	2025-08-31T20:11:58+03:00	Georgi Gerganov		server : enable /slots by default and make it secure (#15630)
c37052ab4d6d1ae73c0e90bc6e560cc6409e1311	5c16b9c87d840e4d5d55fa83c732c6b693346f40	2025-08-31T02:06:43-05:00	Jeff Bolz		vulkan: mul_mat_id coopmat2 optimizations (#15546)
792b44f2ed9668cce7f267ff0ae4950ed9b4a5de	81017865ee444cf49ce0136f2be1e41a0270ff91	2025-08-29T19:25:40+02:00	ExtReMLapin		server : add documentation for `parallel_tool_calls` param (#15647)
e8d99dd0b67f2ecc1e45fca8074a3a18c3e036d2	a8bca68f727844e7dcf24a956003b3c2039ea563	2025-08-28T18:39:31-06:00	Gabe Goodhart		nvidia nemotron nano v2 (nemotronh) (#15507)
c8d0d14e77c3c45df5cbbddde9c2b4c377ec4e7a	84ab83cc0b4b7e769451ee48e4c7d1acef91ef25	2025-08-28T17:09:05+03:00	Georgi Gerganov		kv-cache : fix find_slot to not search for continuous slot (#15638)
64387f6e95434b393ac3df285864692b7fd9c4d2	d35a1e8c41f747548775225973a99507896a8c61	2025-08-28T10:56:41+02:00	Aleksei Nikiforov		gguf-py: byteswapping improvements (#12851)
a6a58d64785cb458ed9de52f391aa38142d38d64	0373486dbc0dccbdcb3b5fdd65759d88cec06196	2025-08-26T21:05:25+05:30	shalinib-ibm		llamafile: PowerPC Sgemm Optimization (#15558)
0115cfb7c1dec0753aecb34f3af4338bf9de8eeb	3a617cb18121a85eed17ac5585788810c5d9c1e0 0d5a470223fc90b6b6807921d68011ff06ae7f9e	2025-08-26T15:19:56+08:00	Yunzhe Jia		Merge branch 'master' into dev
5eff6ec9b1220b599a43b594b1110487ab6aca08	dfd9b5f6c7586c88588f06a644c131bec071a0a1	2025-08-25T17:23:40+02:00	Johannes Gäßler		CUDA: MoE helper in device code, better tile sizes (#15525)
611f419cff11e4952228162a1c44cb35dff2274a	b1afcab804e3281867a5471fbd701e32eb32e512	2025-08-23T13:16:17-05:00	Jeff Bolz		vulkan: optimize rms_norm, and allow the work to spread across multiple SMs (#15281)
b55f06e1aa67fb10e89f53e31bbccf37eb2678ea	0a9b43e507a359ca392c037cf341f55137ad0b69	2025-08-23T14:58:57+08:00	R0CKSTAR		vulkan.Dockerfile: install vulkan SDK using tarball (#15282)
330c3d2d21b55bca5517db7d2eea2ea8f131df4a	e92734d51bcb82cc35f0a6b5a14928f0036b2c90	2025-08-23T01:31:54-05:00	Jeff Bolz		vulkan: optimize mul_mat_id loading row ids into shared memory (#15427)
8ad038c0fdc719ced9fbf921a02cbae9ad79287f	5682a3745f2b653dcb855d5766d8edc318fb3336	2025-08-21T11:06:05+08:00	R0CKSTAR		musa: add GGML_UNUSED_VARS (#15446)
a094f381432d92c4bf92d2d6167284316ba73a62	fb22dd07a639e81c7415e30b146f545f1a2f2caf	2025-08-20T10:17:37+08:00	R0CKSTAR		musa: fix build warnings (#15258)
67f09a3a27db443f9870aac87e163dba0d08131e	6424594c56f4dbd0573455d89a0d89a0ac093d13	2025-08-19T18:33:47+08:00	R0CKSTAR		musa: handle __hgt2_mask, available starting from MUSA SDK rc4.3.0 (#15413)
00f35d509e5367bf19ccaf4b4adddc38db4811c6	6028bf74351d35a06bd98498624f8c2f029f7d1a	2025-08-13T11:09:39+03:00	Georgi Gerganov		ggml : repack block_iq4_nlx8 (#14904)
6028bf74351d35a06bd98498624f8c2f029f7d1a	bc5182272c373267352bc689e5fca276934bea2d	2025-08-13T10:04:46+02:00	Oliver Simons		CUDA: Optimize `reduce_rows_f32` kernel, leading up to 25x perf improvement on kernel-level and 10% perf increase for Gemma3n (#15132)
25ff6f7659f6a5c47d6a73eada5813f0495331f0	be48528b068111304e4a0bb82c028558b5705f05	2025-08-12T10:02:51+08:00	R0CKSTAR		musa: fix failures in test-backend-ops for mul_mat_id op (#15236)
cd3069dfcbeee8e0e96cffb93b0ebb9e595e273a	50e81bdf5db563ab57a9a722b08f96fa8a76c927	2025-08-11T11:21:19+02:00	Daniel Bevenius		kv-cache : log (debug) all streams in find_slot (#15176)
79c1160b073b8148a404f3dd2584be1606dccc66	34c9d765bf173c551398f1e7fa4595019bc53bab	2025-08-09T13:29:43-05:00	David Zhao		cuda: refactored ssm_scan and use CUB (#13291)
34c9d765bf173c551398f1e7fa4595019bc53bab	e54d41befcc1575f4c898c5ff4ef43970cead75f	2025-08-09T20:00:24+08:00	Aman Gupta		CUDA: add attention sinks for tile and wmma (#15178)
9515c6131aecaccc955fdedcfe16c3e030aaefcb	fd1234cb468935ea087d6929b2487926c3afff4b	2025-08-05T16:26:38-07:00	Reese Levine		ggml: WebGPU disable SET_ROWS for now (#15078)
587d0118f50b7e8f4bafbcdd218aefd9da0272e1	5aa1105da24a8dd1661cea3db0582c9b2c2f54d3	2025-08-04T08:52:43-07:00	Reese Levine		ggml: WebGPU backend host improvements and style fixing (#14978)
3025b621d12a6931ff5e9775d4f644719980ad91	ec0b18802c91badd3ff1388ffd09ee163251bd72	2025-08-02T17:20:40+08:00	R0CKSTAR		llama-bench: rename DB table name from test to llama_bench (#15003)
baad94885df512bb24ab01e2b22d1998fce4d00e	ba42794c9ead96ad52311ba1b23eefcbf3d6f63d	2025-08-01T11:50:33+05:30	Srihari-mcw		ggml : Q2k interleaving implementation - x86/x64 SIMD (#14373)
484b2091ce5017901483b5204c07878f171d1441	daf2dd788066b8b239cb7f68210e090c2124c199	2025-08-01T08:47:27+08:00	R0CKSTAR		compare-commits.sh: support both llama-bench and test-backend-ops (#14392)
8a4a85627702b569d7d2810f2de06a4321656e9d	11490b36723d511d75fb601995c79b5c363ba3a2	2025-07-31T19:49:09+08:00	Aman Gupta		Add LLaDA 8b Diffusion model (#14771)
00131d6eaf4df029e1ec84de868c2c5957503007	1e15bfd42c3938506e0da5939bf7f42780965f01	2025-07-30T15:12:02+03:00	Georgi Gerganov		tests : update for LLAMA_SET_ROWS=1 (#14961)
cd1fce6d4f9c191f1c7429cc96f61281c3b63ffc	00fa15fedc79263fa0285e6a3bbb0cfb3e3878a2	2025-07-28T20:32:15+05:30	Akarshan Biswas		SYCL: Add set_rows support for quantized types  (#14883)
9b8f3c6c776d77045ee4f7f13fdf7863dfe59e5b	c7f3169cd523140a288095f2d79befb20a0b73f4	2025-07-26T10:36:02+08:00	R0CKSTAR		musa: fix build warnings (unused variable) (#14869)
793c0d7f46384001738c337d7afa46b45ae32745	ce111d39d666f4a3b6a561ad020f6feb8cc67790	2025-07-25T10:47:39-06:00	Gabe Goodhart		metal: SSM_SCAN performance (#14743)
c12bbde37258c6d053322d1cedd4a9672109ae58	3f4fc97f1d745f1d5d3c853949503136d419e6de	2025-07-25T01:07:26-07:00	Diego Devesa		sched : fix multiple evaluations of the same graph with pipeline parallelism (#14855)
3f4fc97f1d745f1d5d3c853949503136d419e6de	2df255da3cea108de0ae9b302ffdd31674b6d88d	2025-07-25T03:05:37+08:00	R0CKSTAR		musa: upgrade musa sdk to rc4.2.0 (#14498)
cb4a63aad6650c2b536a7578403935388cb2920e	86f5623d904cfd392fdeb14a143097b4074660f6	2025-07-24T11:09:57+01:00	Alberto Cabrera Pérez		sycl: fixed semantics of block offset calculation (#14814)
48b86c4fdb1b1246d9fe17d2d53e3a1c2a0c3245	38d3af1b73302377111e88ddd725257638339286	2025-07-22T07:45:26+08:00	R0CKSTAR		cuda: remove linking to cublasLt (#14790)
6c9ee3b17e19dcc82ab93d52ae46fdd0226d4777	cd465d823c378853f1f6570eebfb77f69c7e1d39	2025-07-21T19:03:19+02:00	rmatif		opencl: add conv2d kernel (#14403)
90083283ec254fa8d33897746dea229aee401b37	d4b91ea7b2da253e1355b503f0fcb7b428ce005d	2025-07-19T12:51:22-04:00	compilade		imatrix : use GGUF to store importance matrices (#9400)
2adf8d83acdb9b1bf58db6c9729ac9dc6847a58b	021cc28bef4dd7d0bf9c91dbbd0803caa6cb15f2	2025-07-18T17:33:41+03:00	Georgi Gerganov		parallel : add option for different RNG seeds (#14757)
086cf81e88fb75287b71ff19c08a206b7bc2e02f	d9b691081c04ec5fb0daa9d2b979f915c142963d	2025-07-17T09:22:11+02:00	Tarek Dakhran		llama : fix parallel processing for lfm2 (#14705)
21c021745d781edf9c44b4972ef6cbbf53b0ecff	b0f0ecc3dce806c68609d375a2b3edc430d8db18	2025-07-16T08:18:51-07:00	Reese Levine		ggml: Add initial WebGPU backend (#14521)
225e7a1438f4ea85eaa7b5ef3ab3b266ee4d9c06	ab140198211385b85eeeb0abd549a4bbe259e10d	2025-07-16T16:35:42+03:00	Georgi Gerganov		llama : add high-throughput mode (#14363)
e4841d24d3485ea4af54f8b65a19ec3123f0ff3c	538cc77f7f44dfa047dba6a06d90c86dda69cf1d	2025-07-16T19:12:22+09:00	Shunta Saito		llama : fix parallel processing for plamo2 (#14716)
cbc68be51d88b1d5531643b926a4b359c3cff131	bdca38376f7e8dd928defe01ce6a16218a64b040	2025-07-15T15:28:53+08:00	R0CKSTAR		cuda: fix build warnings in set-rows.cu (unused variable) (#14687)
7de5c7cab61d4da4387ed9b216f88b96297bcc2d	8eff95544e817704d44bec20f9fc956ce76a33be	2025-07-12T21:31:38+08:00	Aman Gupta		CUDA: add set rows for f32 and f16 (#14551)
b3ad3a0191994d6c47b2bd389d5c7431526ecd2c	98197e5c98388470030d908f355ec5937dcccaaa	2025-07-12T05:12:26-05:00	Jeff Bolz		vulkan: support SET_ROWS (#14587)
0aedae00e6fb48680324a5ac5da9cba0e35de6b5	6bdda13981d6c8189b7dc4f9fb8ecb91c21529f8	2025-07-10T18:20:13-06:00	Gabe Goodhart		model : Granite Four (#13550)
704bb7a71c01dc07c1478b85f6322bf5dfde1eaf	435a6d10d618a015060e45a38c7e9f27f4243316	2025-07-10T13:59:38+05:30	Akarshan Biswas		SYCL: Initial set_rows kernel implementation (#14562)
4a5686da22057867c23bd4a6be941ddc8c51e585	98bab638fb28cf95a5a66dd2d51b40d6c8f6d69a	2025-07-09T14:59:57-04:00	compilade		llama : support Jamba hybrid Transformer-Mamba models (#7531)
6efcd65945a98cf6883cdd9de4c8ccd8c79d219a	699f4392a33f57c3352cf8d60bdc53db7ca235e7	2025-07-08T13:11:42-05:00	Jeff Bolz		vulkan: optimize flash attention split_k_reduce (#14554)
68155c66f0e76680f34442247e589a090add22d3	e1a7059053a9b8958f2d57a21fc46dbc7fb24f8e	2025-07-08T07:58:30+08:00	R0CKSTAR		musa: fix build warnings (unused variable) (#14561)
a0374a67e2924f2e845cdc59dd67d9a44065a89c	ddef99522d1ba74193b7394e803fab8db5c78bae	2025-07-05T02:26:04-05:00	Jeff Bolz		vulkan: Handle updated FA dim2/3 definition (#14518)
a70c8a0c4b4c1606cd9a0ba889ce61aa88610095	9067487c4411efb20400103fcccfdd389c80d428	2025-07-03T10:53:35+03:00	Georgi Gerganov		kv-cache : use ggml_set_rows (#14285)
5d46babdc2d4675d96ebcf23cac098a02f0d30cc	e17991c466ac835b2c71bd813c1ca7ff8dd97b94	2025-07-02T13:10:24-04:00	compilade		llama : initial Mamba-2 support (#9126)
55a1c5a5fdefef808e95aabd3d5563af1068cc80	12a81af45f0dbbab24bd819a15f57c03ceb1be90	2025-07-02T20:34:24+08:00	Aman Gupta		CUDA: add softmax broadcast (#14475)
a0535ffa0d35fccfec3e1a0a3bfc9dbb6054d7c0	bd9c981d7226107f18deb8344c3301450311bb8b	2025-06-29T11:04:10+02:00	Sigbjørn Skjæret		ggml : implement REGLU/GEGLU/SWIGLU ops (#14158)
bd9c981d7226107f18deb8344c3301450311bb8b	27208bf657cfe7262791df473927225e48efe482	2025-06-29T02:43:36-05:00	Jeff Bolz		vulkan: Add fusion support for RMS_NORM+MUL (#14366)
00d5282c7f2a0bb05bb315fb81ca3b0f42cf9f07	566c16fcce44876a167c37f159085afe6f84b28c	2025-06-28T10:17:09-05:00	Jeff Bolz		vulkan: lock accesses of pinned_memory vector (#14333)
b25e92774e2fa4ee3820e458d5cf43f40190f8d2	6609507a910aa7437aaa53fd999447de3947d998	2025-06-28T17:35:41+08:00	Xinpeng Dou		fix async_mode bug (#14432)
5783ae43599400b723b5da0569c1f848419ff3c7	bf5bcd0b857db420235e03639f0a5f218a7f8cf8	2025-06-26T15:50:15+03:00	Georgi Gerganov		metal : batch rows copy in a single threadgroup (#14384)
8f52c6f8abb2ed9ad967cc0bfb90acb073011193	d2f325130c4ed77a3dcff2c23a926587fb08e2cc	2025-06-26T15:10:45+08:00	Yunzhe Jia		Squashed commit of the following:
716301d1b03c31875ec3b24526c48c8b1bd0fd8c	60ef23d6c14d325d83eae5752e5de39ad268e9b0	2025-06-26T12:11:59+08:00	R0CKSTAR		musa: enable fp16 mma (all) and cublas on qy2 (#13842)
73e53dc834c0a2336cd104473af6897197b96277	62af464227dafa1c55e0535bcb24346326748f46	2025-06-24T11:46:25-07:00	lhez		opencl: ref count `ggml_backend_opencl_context` and refactor profiling (#14254)
fa4a9f2a1ccda2573189a9d4995bdf0bceb41156	238005c2dc67426cf678baa2d54c881701693288	2025-06-22T22:16:26+01:00	Ed Addario		quantize : handle user-defined pruning of whole layers (blocks) (#13037)
bb16041caef45cd4348cd6f84906b5dfec7a1f6a	58cba76a9aab728717509d62b67c13afd8dc227a	2025-06-21T08:17:12+02:00	Markus Tavenrath		Add support for VK_EXT_debug_utils to add labels to Vulkan objects. (#13792)
67ae5312e255ae97852a4a216e2245580bfafd72	692e3cdd0a069ab56411b64506a67537d767683e	2025-06-21T08:04:18+03:00	Georgi Gerganov		metal : fix thread-safety (#14300)
8308f98c7fb778e54bf75538f5234d8bd20915e9	6369be07359d03723f38c0a4a014ff0f698a0738	2025-06-20T15:07:21+02:00	Nicolò Scipione		sycl: add usage of enqueue_functions extension (#14244)
e28c1b93fd7d3f8faf9551d962e8a65fe2122e38	d27b3ca1758dfb1718e333d497ef4b68ad109bc2	2025-06-20T04:57:36-07:00	Diego Devesa		cuda : synchronize graph capture and cublas handle destruction (#14288)
edc4a29effe716956fdfd2bc5b9cba2a6d8492f8	ed3290ab34493fd3d2e0f925f816a401da4f2dfd	2025-06-19T00:08:14-05:00	Gabe Goodhart		memory : Hybrid recurrent cache (#13979)
8d947136546773f6410756f37fcc5d3e65b8135d	50d2227953ca9024f04255b4f116d06fcc0db74c	2025-06-19T01:10:26+08:00	Aaron Teo		docs: add s390x build documentation (#14264)
fe9d60e74a6cb71bcaed2029377bfa2872b4abb0	e434e69183fd9e1031f4445002083178c331a28b	2025-06-17T17:48:08+08:00	R0CKSTAR		musa: fix build warning (unused variable) (#14231)
6adc3c3ebc029af058ac950a8e2a825fdf18ecc6	0dbcabde8c006d5cf781ca0fe071c41559572a72	2025-06-16T08:11:43-07:00	Diego Devesa		llama : add thread safety test (#14035)
c89c2d1ab94b11845240b7d3313c87691ea18d88	3555b3004ba7687be3d734acade52a3345758aa4	2025-06-16T00:21:08-06:00	Jeff Bolz		vulkan: mutex around vkQueueSubmit (#14127)
c311ac664d68d10781a3e7b9f02d9d9520837d80	b9912ac570de8945ae9383c9ca8291027bf287dd	2025-06-15T10:08:58+03:00	Georgi Gerganov		cparams : rename LLAMA_MAX_PARALLEL_SEQUENCES to LLAMA_MAX_SEQ (#14188)
c61285e7396c8e526fe7794c19e8d4f1c99bfc51	09cf2c7c655c90e53e100f29b830a788bab0653d	2025-06-13T08:45:37+01:00	Ewan Crawford		SYCL: Bump oneMath commit (#14152)
bd248d4dc7265437e1918dc53ae3f49a0c592e5f	7781e5fe99f2a4fc1c8af0a8488eedac4644cb72	2025-06-11T09:48:52-05:00	Jeff Bolz		vulkan: Better thread-safety for command pools/buffers (#14116)
dad5c44398b78467943ed0a603ea427fe9f6fc62	55f6b9fa6563f6ae49113f9abdc980c12348cc1c	2025-06-10T18:20:14-04:00	compilade		kv-cache : avoid modifying recurrent cells when setting inputs (#13834)
e21d2d4ae26d6c5daf1602a1061aa4f8e722ca57	dc0623fddb661926e0998538895155e4f081ff09	2025-06-09T19:47:39+08:00	Xinpeng Dou		CANN: Simplify the environment variable setting(#13104)
dc0623fddb661926e0998538895155e4f081ff09	87d34b381d5868e75586210ec17b5ef5deddc276	2025-06-09T18:01:17+08:00	R0CKSTAR		webui: fix sidebar being covered by main content (#14082)
228f34c9ceefa3ea4f4d6933edd858121e8106cb	0974ad7a7cd4bca846b15c484ff3be890135a52c	2025-06-07T18:58:20+05:30	Akarshan Biswas		SYCL: Implement few same quantized type copy kernels (#13739)
a7b8d35f780ab3e7639ae5345442fb1ad10f0163	6eba72b71c677ce9aae33c422a6e67008137987a	2025-05-29T13:29:50+03:00	Georgi Gerganov		sync : whisper.cpp (ggml/1250)
fedf034a98378059274e4243d2a370a243335e73	8726392d3d927fe3e504868d3548d694496ee37e	2025-05-27T20:58:46-04:00	Daniel Tang		ggml : Print backtrace on uncaught C++ exceptions (ggml/1232)
c04621711a893cbd09cff6c927cb005bc6749e36	0fc16b42e8793364918e830c234c1f3caec29dae	2025-06-01T11:42:16+03:00	Georgi Gerganov		parallel : fix n_junk == 0 (#13952)
053b1539c02617eff744f89525ee57497c3c1fbe	b3a89c3d9e34c28c5be70d8b687a84775746d4a0	2025-05-31T15:39:19-07:00	Max Krasnyansky		threading: support for GGML_SCHED_PRIO_LOW, update thread info on Windows to avoid throttling (#12995)
12d0188c0dc6146ffde6d277a93f232ccbe699f8	eb3949938e82a128855bc0676220bb2ce6e4228d	2025-05-31T10:24:04+03:00	Georgi Gerganov		kv-cache : refactor + add llama_memory_state_i (#13746)
dd665cc9d4b378626cde11ea0c4c91f514fdc994	df0c0c7d02f951dc639d48fd536f79200460ac83	2025-05-30T19:38:07+03:00	Georgi Gerganov		parallel : increase the variability of the prompt lengths (#13927)
952f3953c1b61cc70e79e536c42ddce6a5ea5ea7	81713121ee56755ba4a147d1a1e3fa248ec31a66	2025-05-27T04:05:18-07:00	Diego Devesa		ggml : allow CUDA graphs when using pipeline parallelism (#13814)
6f180b915c9ed9ec0c240b5dcd64644988fb5e82	03f582ae8fccecff225c30a2802461b44761e822	2025-05-26T21:10:36+05:30	Akarshan Biswas		SYCL: Add non contiguous support in RMS_NORM and NORM kernels (#13611)
de2ef53a4b2c0d703749a309d19fe68fd8f1b9ac	c508256db2de2b032e19c8ed833f4683c827c9a1	2025-05-25T16:34:36+03:00	Georgi Gerganov		kv-cache : rework kv_cell (#13706)
ffd0eae60b7642e942c8245b89642527e36099e6	b775345d788ac16260e7eef49e11fe57ee5677f7	2025-05-24T11:46:19+02:00	Johannes Gäßler		CUDA: fix race condition in FA vector kernels (#13742)
d394a9aedc50a13b7f6373416f7c1ccabfe79c32	6b56a64690a318fcabcd7739ac7e314d44785ea8	2025-05-22T13:54:43+02:00	Nicolò Scipione		sycl : Remove waits from function calls (#13702)
6b56a64690a318fcabcd7739ac7e314d44785ea8	a4e8912dfd4604be1e39bc86ba4c0b02969967ef	2025-05-22T09:24:09+01:00	Ewan Crawford		SYCL: Avoid using with SYCL-Graph for unsupported nodes (#13587)
a4e8912dfd4604be1e39bc86ba4c0b02969967ef	edbf42edfdabb9cea72ae12137570cf48f5d8076	2025-05-22T02:21:45+03:00	Henry Linjamäki		opencl: Add support for multiple devices (#12622)
c76532e7ba128bb097bf6836bf0f5592e1b56b76	2aa777d86d3f7bb80b93e226f1c25e47825f6a83	2025-05-21T19:40:35+03:00	antichristHater		convert : add qwen2vl support for unsloth merges (#13686)
33983057d0f578aca74ba15eccc3de9c267a5ff6	fb1cab201c6c4cd36731f100df74eece2f4706fa	2025-05-21T09:58:49+08:00	R0CKSTAR		musa: Upgrade MUSA SDK version to rc4.0.1 and use mudnn::Unary::IDENTITY op to accelerate D2D memory copy (#13647)
e298d2fbd082a52c0f6ed02729f94e9bf630cf17	f0adb80bf7c2c0d80abb04f4533b5513622d9964	2025-05-20T08:05:46+03:00	Georgi Gerganov		kv-cache : add SWA support (#13194)
8b5e19aea6ce9fe4452598663924373234041440	60aea028b575ab54e0dfbb31db641bb93d2ae8c4	2025-05-18T18:30:13-07:00	Diego Devesa		ggml : fix apple OS check in ggml_print_backtrace (ggml/1229)
60aea028b575ab54e0dfbb31db641bb93d2ae8c4	9c55e5c5c24990dd17fd6c8f4f2159052d2b06f1	2025-05-17T19:06:26-04:00	Daniel Tang		ggml : Fix missing backtrace on Linux (ggml/1228)
518329b2d4434d0683c30b741b4f4cf5734b5f99	2f5a4e1e09567e09be8b84a5f976334de9539d17	2025-05-17T12:58:55+03:00	Georgi Gerganov		parallel : add option for non-shared and larger prompts (#13598)
09232370fc6426aa5dd9be01a8271b9c28f5af3a	7474e00b34629e9cd8b06bc87ad935584ea30f8e	2025-05-11T16:20:39+02:00	Sigbjørn Skjæret		scripts : exit compare-llama-bench.py gracefully when there's nothing to compare (#13451)
0208355f42bdab88a08507ead4a6302790a08323	d2a4ef05c60506ee48e7375eb36f2257de7ab0d2	2025-05-10T22:22:48+02:00	Johannes Gäßler		CUDA: fix race conditions FlashAttention kernels (#13438)
33eff4024084d1f0c8441b79f7208a52fad79858	17512a94d636c4b6c1332370acb3e5af3ca70918	2025-05-09T19:29:37+02:00	Xuan-Son Nguyen		 server : vision support via libmtmd (#12898)
0527771dd80bd18479dfaaa0a98be297fc3592bf	2189fd3b6327a1d17893694125da8edcf74a6468	2025-05-09T17:25:50+08:00	R0CKSTAR		llama-run: add support for downloading models from ModelScope (#13370)
1f73301b63668d61b5f3109489050a27dc3f65be	4773d7a02ffdb05ba9e673ff21ce95351836e33a	2025-05-07T15:48:23+08:00	R0CKSTAR		cuda : remove nrows_x in mul_mat_q_process_tile (#13325)
93c4e23905987949b714b21ae918ff6bfb55fe36	8afbd968182909cf93fb15959fc867b6dd3adb53	2025-05-04T14:16:39+02:00	Johannes Gäßler		CUDA: fix race condition in MMQ stream-k fixup (#13299)
8afbd968182909cf93fb15959fc867b6dd3adb53	8ae5ebcf859b05a2ea3bbd930133a2fe4a89ed3c	2025-05-04T13:58:38+02:00	Johannes Gäßler		CUDA: fix race condition in MMQ ids_dst (#13294)
8efbdadc616fa717c369059b9b388160958d886c	f057808ffadce213f54c1884ab5096e60140f358	2025-05-01T17:32:11-04:00	Justin Santa Barbara		rpc : avoid uninitialized memory in serialize_tensor (#13210)
3e168bede4d27b35656ab8026015b87659ecbec2	ceda28ef8e310a8dee60bf275077a3eedae8e36c	2025-04-30T16:56:24+02:00	Xuan-Son Nguyen		convert : improve model arch handling (#13122)
43ddab6eeeaab5a04fe5a364af0bafb0e4d35065	1831f538f720d1d99fba146f24f0a8e970838cc4	2025-04-28T21:00:20+03:00	Ville Vesilehto		fix(rpc): Improve input validation and error handling (#13069)
5fa9e63be82225fb3249c76f39ddda3e5bdec0a3	a4c340f974f9b7ac0c1aae897aabaa54549a97e5	2025-04-28T12:18:59+02:00	Xuan-Son Nguyen		clip : refactor set input for cgraph + fix qwen2.5vl input (#13136)
a4c340f974f9b7ac0c1aae897aabaa54549a97e5	d0a417f3c7a5a22ef05b3b76d91dbe1d3362bf0c	2025-04-28T15:03:25+05:30	Akarshan Biswas		SYCL: Add all missing unary kernels (#13074)
f0dd6a1926cdb2f4183a937deee40db26ef8f1da	69699be48a6b94570773532850667f1591dc5bbe	2025-04-28T15:33:28+08:00	R0CKSTAR		musa: fix typo in cc control (#13144)
e291450b7602d7a36239e4ceeece37625f838373	59e991c23cc44cb5fb657e7b9358cac21fb79828	2025-04-27T19:22:49+08:00	R0CKSTAR		musa: fix build warning (#13129)
ca2bb89eac2097ab4620448737e58af8452e444b	2d451c80590b9ac250322769ac13d3b4870dbcf7	2025-04-27T16:10:34+08:00	HimariO		clip : Add Qwen2.5VL support (#12402)
2cca6c01e46d2fc1124d15730273ed2acdad1016	658987cfc9d752dca7758987390d5fb1a7a0a54a	2025-04-23T10:32:49+03:00	Radoslav Gerganov		rpc : add command line option for number of threads for the CPU backend (#13060)
7a395f67a7a02bb361d944b816d6e933889e28e1	971f245b3b5f3f55991bb779cb541b00f82eea1d	2025-04-17T20:34:16+08:00	hipudding		CANN: Add support for async operator submission (#12864)
b0c75ac9f93322b45e3766e149417334b4fd1ed9	d6d2c2ab8c8865784ba9fef37f2b2de3f2134d33	2025-04-15T10:04:24+08:00	Xinpeng Dou		CANN: Optimize CANN buffer pool memory management (#12875)
8ac9f5d765f2b1b7f821873618c0cff68ca59bc8	12e9158f25cb47e97ca9b8083e1ec7415f262260	2025-04-11T15:26:17+08:00	R0CKSTAR		ci : Replace freediskspace to free_disk_space in docker.yml (#12861)
64eda5deb9859e87a020e56bab5d2f9ca956f1de	fe5b78c89670b2f37ecb216306bed3e677b49d9f	2025-04-10T17:24:44+02:00	Xuan-Son Nguyen		convert : ability to lazy-load safetensors remotely without downloading to disk (#12820)
d9a63b2f2e91cdcb0eda211b7f49fadcdea0f664	8ed71242f464dc0a3fb3cffcfe064e55bdec72f9	2025-04-09T17:22:30+08:00	R0CKSTAR		musa: enable freediskspace for docker image build (#12839)
0090950f679475c5ecaac2f7bca5049cca96492b	7ecd780b1a1d5214b8d04c25ebfc194d310816ed	2025-04-09T00:25:08-05:00	Jeff Bolz		vulkan: In coopmat2 mmq, load q4_k/q5_k scales through shared memory (#12833)
78a1ba0a4f2bfed5b8b8e312592143d22e531698	2dabf759e7c8c827d38cdd18d0792070e6a4f4e1	2025-04-08T18:37:06+02:00	Xuan-Son Nguyen		server : fix thread.join() on exit (#12831)
916c83bfe7f8b08ada609c3b8e583cf5301e594b	0c74b04376b0b9efc096480fe10f866afc8d7c1c	2025-04-06T21:23:54+08:00	R0CKSTAR		musa: fix compilation warnings in mp_22/31 (#12780)
5f696e88e0eb490c8028421d3c5419df9dcf5385	193c3e03a63ccda3ac3d6a2999e41e6d1414fe23	2025-04-03T19:51:35+08:00	R0CKSTAR		sync : minja (inclusionAI/Ling) and update tests (#12699)
a10b36c91a091f4606710fba4e9327fd71e0e738	83a88bd6affbe148a622ac730952ac5b8b585979	2025-04-02T14:32:59+03:00	Georgi Gerganov		llama : refactor kv cache guard (#12695)
a6f32f0b3437ac79827ffb55521cb839343324ab	2bb3597e426ce092c3e10ae0bdd876963765e6ed	2025-04-01T19:12:53+08:00	R0CKSTAR		Fix clang warning in gguf_check_reserved_keys (#12686)
c80a7759dab10657b9b6c3e87eef988a133b9b6a	250d7953e829ff0e16074510fef0e7cec696461b	2025-03-31T18:40:56+02:00	Daniel Bevenius		vocab : add special infill tokens for CodeLlama (#11850)
6c02a032fa21d69e881ef9a5c94ba28ebaf1d749	f52d59d771dc231fc2ac39adacf157ddefc97730	2025-03-31T14:55:24+05:30	Akarshan Biswas		SYCL: Remove misleading ggml_sycl_op_flatten function (#12387)
492d7f1ff77e116e44962b832cc3db24cfc46d24	d3f1f0acfbf8aaafd2ce494309a10a7cc92042c8	2025-03-30T16:59:38+08:00	R0CKSTAR		musa: fix all warnings, re-enable `-DLLAMA_FATAL_WARNINGS=ON` in ci and update doc (#12611)
c7b43ab60855f752ae79937fb93d561bc30b69a4	24feaec05792b972d5ff3e2b12d9237ebd50d1ac	2025-03-27T12:21:47+05:30	amritahs-ibm		llamafile : ppc64le MMA implementation for Q4_0. (#12489)
f17a3bb4e8b0aa24c0f86636d234aca7dc2cfa01	bd40678df768ab189b26b8b03e3de4062e3b71a3	2025-03-27T07:16:00+05:30	Akarshan Biswas		SYCL: implement memset ggml backend buffer interface (#12580)
fd7855f8f522ab26681b25ae506ff99c654e2dd5	53af4dba425768fd507ce6b45873cfbb3df2295f	2025-03-26T15:09:48+08:00	R0CKSTAR		doc: [MUSA] minor changes (#12583)
3cd3a395323fa9cdf6ecfa1fea290bf228d4e856	2d77d88e70d017cd82c3f1a4517e3102e2028ac4	2025-03-25T15:45:08+08:00	R0CKSTAR		ci: [MUSA] add CI and update doc (#12562)
7ea75035b67f44c22ed7039967f718011fd35ce5	c54f6b7988b63714b87b0390e01a1e69aee79a12	2025-03-24T18:28:34+08:00	R0CKSTAR		CUDA: Fix clang warnings (#12540)
fac63a3d786b2a0f97876c30add02cb525a9648e	eddfb438502bd5d1014d63a812e9b6d03d326f8c	2025-03-22T17:11:37+08:00	R0CKSTAR		musa: refine compute capability (#12493)
3d82dbcbce2c677fc35fbf99574ccd107d95a1f8	732b5fbf5e7f9cf069942f0c5850ee959ef321ba	2025-03-20T17:05:34+05:30	Srihari-mcw		ggml : block interleaving support for Q4_K quantization for x86 AVX2 architecture (#12332)
517b5ddbf002b91fd6d6daf5d8db8c88a0173039	a9b59288e222f39fc0311dc66944ed5a86c815fa	2025-03-20T01:22:06+05:30	Gaurav Garg		CUDA: Improve flash decoding kernel GPU occupancy for BS=1 case (#12183)
d84635b1b085d54d6a21924e6171688d6e3dfb46	75422e8bc42646005be0754f7aa438b97a5e777e	2025-03-18T12:54:55-07:00	lhez		opencl: improve profiling (#12442)
bb115d2bf7ed2cdd7dccd7ae74cc9cfe4b0adb71	29fff308c704c1c752cdb5153361e545e2bac09d	2025-03-19T02:28:26+08:00	R0CKSTAR		musa: override warp_size of musa device to 32 (#12445)
7dfad387e3f6ac98d383ded2d175eb59736a3993	60c902926c928f9c2cd6390ce411876f92feeaf3	2025-03-18T07:27:50+08:00	Molly Sophia		llama: Add support for RWKV v7 architecture (#12412)
9f2250ba722738ec0e6ab684636268a79160c854	774973b8f3d5e375b0b74d58638eeb1817e950a8	2025-03-14T16:41:20Z	Eric Curtin		Add CLI arg to llama-run to adjust the number of threads used (#12370)
10f2e81809bbb69ecfe64fc8b4686285f84b0c07	ba7654380a3c7c1b5ae154bea19134a3a9417a1e	2025-03-11T20:16:03+01:00	uvos		CUDA/HIP: refractor mmqv to unify the calculation of nwarps and rows per block between host and device code. (#12177)
251364549fe4a78d4e2e66f1adfaf2bd53041d2c	8acdacb3ea00697477eb019efbb6fc183371055c	2025-03-11T01:18:25+08:00	R0CKSTAR		musa: support new arch mp_31 and update doc (#12296)
05e6f5aad0a4bb48fb2775f2a78505540fdbc47d	673cfef9aa8daad09966208909f346eb996628a4	2025-02-28T13:06:12+05:30	Prashant Vithule		ggml: aarch64: implement SVE kernels for q2_k_q8_k vector dot (#12064)
f777a73e18c218848bca0748581c043987348a5d	af7747c95ae71a5db4184947f837179e82c70b77	2025-02-23T13:14:32Z	Eric Curtin		Some llama-run cleanups (#11973)
0b3863ff9576872855b0265da2cab2ce7e25c4d9	ee02ad02c56ff36a5edd22d8617ab3f9546ce7fe	2025-02-21T15:46:23+08:00	Bodhi		MUSA: support ARM64 and enable dp4a .etc (#11843)
c5d91a74006c49376590ef2b67420bc7ce206397	4806498bf15ef767de3776b00856e56c373dd1b1	2025-02-20T14:06:51+01:00	Charles Xu		ggml-cpu: Add CPU backend support for KleidiAI library (#11390)
73e2ed3ce3492d3ed70193dd09ae8aa44779651d	f7b1116af102bcac450c1a522e9c59db241c6767	2025-02-17T14:03:24+01:00	Johannes Gäßler		CUDA: use async data loading for FlashAttention (#11894)
c48f630d1c1942fce08aa7cb18a53ace443cd611	bd6e55bfd3f3af3e5705cf87a10e5178cef7c3c1	2025-02-13T14:46:59+01:00	Daniel Bevenius		llama : add --completion-bash option (#11846)
bd6e55bfd3f3af3e5705cf87a10e5178cef7c3c1	c7f460ab882065ec5696e3d51e24dbf67b539287	2025-02-13T20:28:18+08:00	R0CKSTAR		musa: bump MUSA SDK version to rc3.1.1  (#11822)
e4376270d971cff7992bdb6c5412a739195b1459	3e693197724c31d53a9b69018c2f1bd0b93ebab2	2025-02-13T01:25:34-05:00	Oleksandr Kuvshynov		llama.cpp: fix warning message (#11839)
a394039db004c6ee00098250d160b5aa018c2314	be3bbd62153820ff6d358c817360927f429105c4	2025-02-13T01:02:38+01:00	Diego Devesa		ggml-cpu : add chunking support to mul_mat_id (#11666)
748ee9fe9312acb8755ee4bf46fd9de2e6a45f29	198b1ec611a2c551ea40e5b9c0b862f37555a4cc	2025-02-12T13:57:33Z	Richard		ggml : fix multi-threaded clamp_f32 (#11824)
4d3465c5aeca8be29cac77f1535c35f4fb274eca	d80be897acdca45f8f7ea2e52c930b1d0e738a14	2025-02-08T15:30:53+01:00	Karol Kontny		ggml: Fix data race in ggml threadpool (#11736)
2fb3c32a1634488a5265d1304ab37628eeb5480d	9ab42dc722ad19a12af80f38a06474d498f96da3	2025-02-06T17:32:29+01:00	Xuan-Son Nguyen		server : (webui) migrate project to ReactJS with typescript (#11688)
3ec9fd4b77b6aca03a3c2bf678eae3f9517d6904	3962fc1a7956dd0afacfbce10fd1a3ffd3ad857e	2025-02-05T02:18:38+08:00	fxzjshm		HIP: force max threads per block to be 1024 (#11621)
a83f528688324a21484a97af1d1be5e1bc8d4c8e	b1bcd309fc8ac929cbd4a6207b3a19886bda031f	2025-01-31T14:15:25Z	Olivier Chafik		`tool-call`: fix llama 3.x and functionary 3.2, play nice w/ pydantic_ai package, update readme (#11539)
4314e56c4f8c5091f45732f39bd94c0c6c323798	496e5bf46bab757d05e18227cb4e17055ae42f42	2025-01-30T11:05:00+01:00	Daniel Bevenius		server : use lambda instead of std::bind (#11507)
e0449763a4f335cca374254a72892141f41eaa59	eb7cf15a808d4d7a71eef89cc6a9b96fe82989dc	2025-01-30T05:48:14+01:00	Daniel Bevenius		server : update json snippets in README.md [no ci] (#11492)
1a0e87d29152cb9d4ce13b4ad64b0382c9ba1ab6	d2e518e9b4231077013ffb0e5c4cc24a0c4c1b7d	2025-01-23T11:59:08-08:00	William Tambellini		ggml : add option to not print stack on abort (ggml/1081)
d2e518e9b4231077013ffb0e5c4cc24a0c4c1b7d	b636228c0ad0db95bf73008094c6145a05615cf6	2025-01-17T21:29:08+09:00	issixx		ggml-cpu : fix ggml_graph_compute_thread did not terminate on abort. (ggml/1065)
5245729e3317064959faefc5e5cbc63f4e9cfbea	6152129d05870cb38162c422c6ba80434e021e9f	2025-01-23T01:01:17-06:00	Jeff Bolz		vulkan: fix diag_mask_inf (#11323)
6171c9d25820ccf676b243c172868819d882848f	e28245f35f2faaf249dd352998b3693a8cc28c51	2025-01-21T13:18:51Z	Olivier Chafik		Add Jinja template support (#11016)
adc5dd92e8aea98f5e7ac84f6e1bc15de35130b5	f11cfdfd7fe29436fce512d934c2ff6b94bd89d2	2025-01-15T19:50:13Z	Eve		vulkan: scale caching for k quants + misc fixes (#11081)
6369f867a410416239d9f20ec27c2b1d6a9fee52	47182dd03fe04a4ffda5d7f4c8a109ae0056cf56	2025-01-06T10:28:17+01:00	Daniel Bevenius		llama : rename missed batch params/vars to ubatch (#10059)
0da5d860266c6928b8c9408efbd264ae59fedda6	a45433ba209ee0b33d02c7dc4c31f29894ad83a6	2025-01-02T15:05:18+01:00	Xuan Son Nguyen		server : allow using LoRA adapters per-request (#10994)
a813badbbdf0d38705f249df7a0c99af5cdee678	fdd21889123bec62b1db3b2fc22b5a4abab32174	2024-12-29T03:16:34-06:00	Jeff Bolz		vulkan: im2col and matmul optimizations for stable diffusion (#10942)
227d7c5a7f4cc7734fde8a4ef4382d613486d3c8	7b1ec53f56bb72c49e4c8434157895f94d3709c2	2024-12-17T09:52:09+01:00	Xuan Son Nguyen		server : (UI) fix missing async generator on safari (#10857)
7b1ec53f56bb72c49e4c8434157895f94d3709c2	160bc039c862c10b9938269a90ddb09d794a7b0d	2024-12-17T05:52:55Z	Eve		vulkan: bugfixes for small subgroup size systems + llvmpipe test (#10809)
5478bbcd173e7027af7689493c7421719f5c43df	b5ae1ddff93403300fc79c7ab5ee73b8cfbb3457	2024-12-15T05:55:54-06:00	Vinesh Janarthanan		server: (UI) add syntax highlighting and latex math rendering (#10808)
a76c56fa1a3d27467eb97468d8c3b2fe1243b61a	c27ac678dd393af0da9b8acf10266e760c8a0912	2024-12-13T12:23:52-08:00	lhez		Introducing experimental OpenCL backend with support for Qualcomm Adreno GPUs (#10693)
11e07fd63bac1cb642380a7a3eac03fd8703e948	4601a8bb6784d2ab8b4b605354b51979fbeea1d3	2024-12-13T18:23:50+01:00	Corentin REGAL		fix: graceful shutdown for Docker images (#10815)
83ed24a97b500ccdb32b90b94e6f9621ad8db79e	d583cd03f663701858ba152a334cbb467fabe342	2024-12-13T12:12:15+05:30	Akarshan Biswas		SYCL: Reduce most of the compiler warnings (#10748)
3573fa8e7b7f0865638b52b4e9b4d2006f0558a2	d9c3ba2b7749c00df477599aa141a98b4521aa2c	2024-12-07T20:21:09+01:00	Xuan Son Nguyen		server : (refactor) no more json in server_task input (#10691)
ce4a7b849388b67d45ad420bdd82d5efcd55647a	19d8762ab61df8286367588a80b9c7db4cb568db	2024-12-07T18:02:05+02:00	Georgi Gerganov		server : various fixes (#10704)
e9e661bd59364e5d4fce035834b6cadcadf8c2ef	efb6ae963031709fc331e6e48cc4606ac8f9c3a7	2024-12-03T21:11:43+08:00	mahorozte		CUDA: remove unnecessary warp reduce in FA (ggml/1032)
efb6ae963031709fc331e6e48cc4606ac8f9c3a7	667d70d1704dfa6977505f5d01d4638669b90dce	2024-12-02T19:27:24+01:00	PAB		feat: add `GGML_UNARY_OP_ARGMAX` Metal kernel (ggml/1019)
0533e7fb3842a523f64dc533bd7bd7147ec2c63a	7cc2d2c88908fc92b97b28acafb82f7d6e425b85	2024-11-30T07:00:02Z	Eve		vulkan: Dynamic subgroup size support for Q6_K mat_vec (#10536)
6c595676899013102fdb0aa4b06a49954300c94a	890719311b6535e572f15965c6d7ec4ac2537f60	2024-11-28T19:17:49+01:00	Xuan Son Nguyen		server : (tests) don't use thread for capturing stdout/stderr, bump openai client library (#10568)
249cd93da3df9c8fa78869b0522526d1625aca91	904109ed0d97c9b656a5e8bf612925f739bb8166	2024-11-27T00:00:41+08:00	R0CKSTAR		mtgpu: Add MUSA_DOCKER_ARCH in Dockerfiles && update cmake and make (#10516)
45abe0f74ee281aea6e5283c1e738061256cfcae	0bbd2262a3263f37385297b30de37941836e57f7	2024-11-26T16:20:18+01:00	Xuan Son Nguyen		server : replace behave with pytest (#10416)
84e1c33cde9e0a7aafcda2d4f21ba51c300482d7	811872a59daefb25fc0c4326bcb6d8ae893c2f7c	2024-11-26T13:36:40+02:00	Georgi Gerganov		server : fix parallel speculative decoding (#10513)
9ca2e677626fce759d5d95c407c03677b9c87a26	5931c1f233c616083d64e41a228249d58e039aa5	2024-11-25T16:31:38+02:00	Georgi Gerganov		server : add speculative decoding support (#10455)
cce5a9007572c6e9fa522296b77571d2e5071357	dc39012cbaf8752fabecaeb60af78ccdd1dfb73b	2024-11-24T18:03:25+02:00	Georgi Gerganov		flake.lock: Update (#10470)
2eb76b2a5e4ea395b971a419c95b473ab6f253e4	9b75f03cd2ec9cc482084049d87a0f08f9f01517	2024-11-18T16:08:20+02:00	Georgi Gerganov		flake.lock: Update (#10346)
cf32a9b93ad859ea592c31785b6bd3b4b2121463	a43178299c2f74f14bcfc659bfd9fd32d931d1f4	2024-11-17T11:23:01+02:00	Georgi Gerganov		metal : refactor kernel args into structs (#10238)
f0204a0ec70d50ca60e07bc0096ec1d6508ab0c7	57f8355b29a8c7dfcd1fb6094758ad85644f8535	2024-11-15T19:47:25+08:00	R0CKSTAR		ci: build test musa with cmake (#10298)
9901068ac78838745e604fffb4601d315a610456	231f9360d94446cd083b6b116f63991b1328c484	2024-11-15T05:48:49-04:00	Xuan Son Nguyen		server : (web UI) add copy button for code block, fix api key (#10242)
ae8de6d50a09d49545e0afab2e50cc4acfb280e2	4a8ccb37ad9c9027cbcfd5548c19cdffe48d5197	2024-11-14T18:04:35+01:00	Diego Devesa		ggml : build backends as libraries (#10256)
fb4a0ec0833c71cff5a1a367ba375447ce6106eb	5ea926dad7f62ebccff7b24784bd1e01a06d13ae	2024-11-13T20:00:35+02:00	Michael Podvitskiy		llama : propagate the results of `graph_compute` (#9525)
54ef9cfc726a799e6f454ac22c4815d037716eda	b0cefea58a20020754b7431e3c725625274372a5	2024-11-11T11:13:51-06:00	Jeff Bolz		vulkan: Throttle the number of shader compiles during the build step. (#10222)
4b3a9212b602be3d4e2e3ca26efd796cef13c55e	505f33274d60676320216b662a97672a76ec600e	2024-11-10T21:45:25+02:00	Georgi Gerganov		flake.lock: Update (#10243)
3bcd40b3c593d14261fb2abfabad3c0fb5b9e318	5c333e014059122245c318e7ed4ec27d1085573c	2024-11-07T18:19:10+11:00	Zhiyuan Li		Optimize RWKV6 Operator Naming and Implement Multi-core CPU/ SYCL Acceleration (#10133)
b11f9ba9b8ce319f04b88afe40d264e6b7f4ba46	94d8cb8be13b7c4d04eeca5a2b956b9148e6f222	2024-11-06T13:29:01+02:00	Georgi Gerganov		server : remove hack for extra parallel slot (#10187)
9e0ecfb697d297355e43c20559d29bcc71beb0c3	6a066b9978533e2ab9890b7f4f8c0262d91798b3	2024-11-04T16:33:29+01:00	Xuan Son Nguyen		server : clarify /slots endpoint, add is_processing (#10162)
1839f69130151ceeac4d01c0ef8964e1fb43bba6	9830b6923b61f1e652a35afeac77aa5f886dad09	2024-11-03T15:14:15+02:00	Georgi Gerganov		flake.lock: Update (#10146)
42cadc74bda60afafb45b71b1a39d150ede0ed4d	45950415ed985830c59bf42cf9c9216b20cf08ef	2024-11-02T16:34:56Z	sasha0552		server : fix slot selection by lru (#10126)
d865d1478cd4e403f82d793c2afcd0f943412f05	1804adb0cfee4811eaf633741503d683a46e4c77	2024-11-01T13:33:14Z	sasha0552		server : fix smart selection of available slot (#10120)
07028f9d74d895da2ca4a1956624e3f07e04e620	524afeec9dad7d765ce91f5cf30c73703867cb47	2024-10-28T17:41:24+02:00	Georgi Gerganov		flake.lock: Update (#10063)
524afeec9dad7d765ce91f5cf30c73703867cb47	8125e6cbfcf2b3b9066e4d923aca9295526730f5	2024-10-28T17:02:48+08:00	R0CKSTAR		musa: workaround for Guilty Lockup in cleaning src0 (#10042)
8c60a8a46261ffb92b6d23a78acfac2fcb6fe233	9e4a2563eadf34e9432d248224d4f43e8495e8fe	2024-10-23T14:34:00-04:00	bssrdf		increase cuda_cpy block size (ggml/996)
bc5ba007b2c83ac95875e68724dabfc12159fc61	958367bf530d943a902afa1ce1c342476098576b	2024-10-25T10:13:46+03:00	Georgi Gerganov		server : check that the prompt fits in the slot's context (#10030)
958367bf530d943a902afa1ce1c342476098576b	40f2555797f97314de749873cdc29dc102be66e2	2024-10-24T21:51:22+02:00	Xuan Son Nguyen		server : refactor slot input data, move tokenizer to HTTP thread (#10023)
873279b1592e433c4d9eb5065091cc98473c7bee	c8c07d658a6cefc5a50cfdf6be7d726503612303	2024-10-20T00:22:59Z	github-actions[bot]		flake.lock: Update
60ce97c9d809f4b040e90b597468b839df5728d0	8901755ba328643c9ab071c20e1939ea52951a0e	2024-10-18T13:34:36+08:00	Ma Mingfei		add amx kernel for gemm (#8998)
fbc98b748e7b075e327bcf13237057f647678049	dcdd535302fc9702a4709be25f56540d65163a44	2024-10-15T15:54:55+05:00	MaggotHATE		sampling : add XTC sampler (#9742)
92be9f12164f18ce845a5bab60cefa5f7fec6836	edc265661cd707327297b6ec4d83423c43cb50a5	2024-10-13T06:11:26+03:00	Georgi Gerganov		flake.lock: Update (#9870)
1bde94dd024b632f98428f4bf2ce483295130779	95c76e8e92ecc93f784b185eafae36a0e7ad2fa3	2024-10-12T16:06:31+03:00	Georgi Gerganov		server : remove self-extend features (#9860)
11ac9800aff532715a5bc7991062c68ba3472e6e	943d20b4111c746bcd9dbc7e4771de313b08b50c	2024-10-12T08:21:51+03:00	Georgi Gerganov		llama : improve infill support and special token detection (#9798)
943d20b4111c746bcd9dbc7e4771de313b08b50c	96776405a17034dcfd53d3ddf5d142d34bdbb657	2024-10-12T13:09:53+08:00	R0CKSTAR		musa : update doc (#9856)
cf8e0a3bb9c0e93e371773b282054cdbbb231038	c7499c557cc1efafaf0a6bc12963c39826299703	2024-10-11T02:10:37+08:00	R0CKSTAR		musa: add docker image support (#9685)
6279dac039ddeb6d5ebd125a6274fd3c37a77ba8	d5ac8cf2f2e30459489e6721a17d15b92a0c42a6	2024-10-07T19:35:42+03:00	Georgi Gerganov		flake.lock: Update (#9753)
96b69121033d2b6b951d1b6b1b43f8b4f97dac99	d5cb86844f26f600c48bf3643738ea68138f961d	2024-10-07T13:26:31+01:00	Paul Tsochantaris		metal : single allocation of encode_async block (#9747)
ace4f4be37abed4801fbd54a94cf38a7ae462416	8277a817f18967581b02b2248989d773e8e99998	2024-09-30T17:48:49+03:00	Georgi Gerganov		flake.lock: Update (#9680)
641002fba8d2a0c0269027e23d2ef58e90546028	0de8b203f1d31cf5ee0d2a3560a0ad78d44f4d4c	2024-09-29T11:50:17-05:00	Jeff Bolz		vulkan : multithread pipeline creation (ggml/963)
544f409b4bd8fc98a3e87820f0ac934e00402de7	6084bfb261b03f812de2255b05b6b5bb8d1c7171	2024-09-26T08:59:42+02:00	Salvatore Mesoraca		vulkan : argsort barriers must be under uniform control flow (ggml/951)
6084bfb261b03f812de2255b05b6b5bb8d1c7171	faac0bae265449fd988c57bf894018edc36fbe1e	2024-09-24T13:23:59+03:00	Georgi Gerganov		ggml : fix GGML_MAX_N_THREADS + improve formatting (ggml/969)
1b2f992cd2cff0b69e5abe78bb8888d51ed19d67	739842703e32cd43443c45e0b4f6647cc4e6b3d6	2024-09-28T14:32:46+02:00	slaren		test-backend-ops : use flops for some performance tests (#9657)
739842703e32cd43443c45e0b4f6647cc4e6b3d6	6102037bbb55521880ae78a6ee6c2a0c00c901df	2024-09-28T15:13:21+03:00	Georgi Gerganov		llama : add comment about thread-safety [no ci] (#9449)
7691654c68aa5316108f881136c59f815ccb6809	ea9c32be71b91b42ecc538bd902e93cbb5fb36cb	2024-09-26T09:27:40+08:00	R0CKSTAR		mtgpu: enable VMM (#9597)
3d6bf6919f7b10726421779cd344f2da05421c68	904837e0cb2f5f01bf5d5901b7aa57a026860ae4	2024-09-25T01:06:52-06:00	Gabe Goodhart		llama : add IBM Granite MoE architecture (#9438)
c087b6f11d3385f4293b6841ebfb755063479490	116efee0eef09d8c3c4c60b52fa01b56ddeb432c	2024-09-23T21:18:48-07:00	Max Krasnyansky		threads: fix msvc build without openmp (#9615)
f0c7b5edf82aa200656fd88c11ae3a805d7130bf	1d48e98e4f3316bd2f6b187d288c7b6cb88d5cb3	2024-09-23T11:42:43-07:00	Max Krasnyansky		threads: improve ggml_barrier scaling with large number of threads (#9598)
f3979df762b75a2b1c0f622a2cd15d1bc60f037f	1e7b9299c6ccb5bbc55d3db7cfa9b51f3ab09b59	2024-09-23T18:43:40+03:00	Georgi Gerganov		flake.lock: Update (#9586)
c35e586ea57221844442c65a1172498c54971cb0	912c331d3dba3a079815844208dc36164baa8cc7	2024-09-22T22:55:49+08:00	R0CKSTAR		musa: enable building fat binaries, enable unified memory, and disable Flash Attention on QY1 (MTT S80) (#9526)
424c5d00a9b97dd5559635872db9b57f87c23b02	a6809c6a2e8163cba296d4cc0c5cd4bbc8073638	2024-09-20T19:04:44+03:00	Johannes Gäßler		ggml/examples: add backend support for numerical optimization (ggml/949)
a6809c6a2e8163cba296d4cc0c5cd4bbc8073638	5cb12f68395a5ec00b357e408883ce124a11dcdb	2024-09-08T11:10:43+03:00	Georgi Gerganov		examples : add null threadpool args where needed (ggml/0)
64c6af3195c3cd4aa3328a1282d29cd2635c34c9	0d2f22e45c3c3b6f8222acb6284d0c8c93443ba1	2024-09-18T19:13:08+02:00	slaren		ggml : fix n_threads_cur initialization with one thread (#9538)
8b836ae731bbb2c5640bc47df5b0a78ffcb129cb	8344ef58f8cdb8ebd6faf4463ca32ae91c374c81	2024-09-17T09:35:38-04:00	Bert Wagner		arg : add env variable for parallel (#9513)
0226613853133c081b55bb892a41bb5eacc0bc94	503147a9f9d195d6a14e7c998df23b6eb61f2bae	2024-09-17T01:19:46-07:00	Max Krasnyansky		threadpool : skip polling for unused threads (#9461)
5c3d0f1824714e9a97fc9b06e046eefcb6ecc721	0aadac10c7dd704f8285ddf5a63d6f764cb340aa	2024-09-16T06:48:24Z	Eve		ggml : IQ4_NL sgemm + Q4_0 AVX optimization (#9422)
90a2fff0e7f80c6fea3fc6cf9a7b482744f3f164	6262d13e0b2da91f230129a93a996609a2f5a2f2	2024-09-16T05:14:23+03:00	Georgi Gerganov		flake.lock: Update (#9488)
befaf1197fa447f61714de041828852a270659d2	feff4aa8461da7c432d144c11da4802e41fef3cf	2024-09-14T09:50:12+02:00	Daniel Bevenius		llama : make cell_id const in inp_s_mask block (#9470)
5af118efdaf1098798a06b24fd8a557760e99631	d2b496bff4f353a6429f8e833448f071bd237ba7	2024-09-11T10:22:40+02:00	Johannes Gäßler		CUDA: fix --split-mode row race condition (#9413)
b34e02348064c2f0cef1f89b44d9bee4eb15b9e7	51b603863627c4074e77b7e556e18ece86bdf9a3	2024-09-11T09:46:55+08:00	R0CKSTAR		musa: remove Clang builtins mapping (#9421)
cb9c933eb2a0d2b514556bdcb934b56dfe5d6771	6cd4e034442f71718563e600070c2b6fc389e100	2024-09-11T01:46:59+03:00	Georgi Gerganov		flake.lock: Update (#9360)
5fac4d57643b1de8e9ab746f14d2fc4e319ae0c2	5fb5e24811cb01d48b482c15a974bfbd9f433e1d	2024-09-09T21:07:18+05:30	Prashant Vithule		ggml : vector length agnostic SVE support (#9290)
efe6a83e3046a94cda38c8be5a7801eadc21d78d	fbb7fcffbcfc50009c275e75982b1603a6cb6b80	2024-08-28T10:23:02+02:00	Salvatore Mesoraca		ggml : fix cont with transposed tensors when one dimension is 1 (ggml/934)
9b2c24c0993487d3b34a873980e292da571481f3	134bc38ecf3e2c5460581badce289a1ffa680453	2024-09-06T23:21:29+02:00	Xuan Son Nguyen		server : simplify state machine for slot (#9283)
4a1411b4f17b6569dc0a067e5914dcc0f738b370	8ebe8ddebd68526757c631cd019de009697c63c2	2024-09-06T14:06:04+02:00	Xuan Son Nguyen		server : fix missing lock (#9334)
7605ae7daf31c02211bcfec2f46635ef6ec4b98a	8962422b1c6f9b8b15f5aeaea42600bcc2d44177	2024-09-04T02:36:43+03:00	Georgi Gerganov		flake.lock: Update (#9261)
f771d064a95d212ddadea452ad0cc1e42b2c3db3	6e7d133a5f9409dd257fad90d7f320721b07a1b2	2024-09-02T08:25:30-07:00	yuri@FreeBSD		ggml : add pthread includes on FreeBSD (#9258)
6e7d133a5f9409dd257fad90d7f320721b07a1b2	b60074f1c26d8354053a952844ef3f7ebefd8803	2024-09-02T17:11:51+02:00	Xuan Son Nguyen		server : refactor multitask handling (#9274)
8f1d81a0b6f50b9bad72db0b6fcd299ad9ecd48c	a47667cff41f5a198eb791974e0afcc1cddd3229	2024-09-01T22:38:17+08:00	Molly Sophia		llama : support RWKV v6 models (#8980)
42c76d1358021ccdbe8ba89109c143dd7ae166df	9f7d4bcf5c27d37b0c7da82eeaf9c1499510554b	2024-08-29T19:20:53-04:00	Faisal Zaghloul		Threadpool: take 2 (#8672)
1d1ccce67613674c75c9c7e3fa4c1e24e428ba48	9fe94ccac92693d4ae1bc283ff0574e8b3f4e765	2024-08-29T07:28:14+03:00	Georgi Gerganov		flake.lock: Update (#9162)
a1631e53f6763e17da522ba219b030d8932900bd	fc54ef0d1c138133a01933296d50a36a1ab64735	2024-08-21T17:58:11-04:00	compilade		llama : simplify Mamba with advanced batch splits (#8526)
8455340b874aa90d5f70104c99ed2778d9e31c36	2f3c1466ff46a2413b0e363a5005c46538186ee6	2024-08-21T09:32:58+02:00	Daniel Bevenius		llama : std::move llm_bigram_bpe from work_queue (#9062)
554b049068de24201d19dde2fa83e35389d4585d	2339a0be1c8e31fcf4531427183b94f2ef019e56	2024-08-18T17:43:32+03:00	Georgi Gerganov		flake.lock: Update (#9068)
8b3befc0e2ed8fb18b903735831496b8b0c80949	d565bb2fd5a2a58b9924a7a34e77a87c78c52137	2024-08-16T17:19:05+02:00	Xuan Son Nguyen		server : refactor middleware and /health endpoint (#9056)
5fd89a70ead34d1a17015ddecad05aaa2490ca46	98a532d474c73d3494a5353024cb6a4fbbabbb35	2024-08-14T18:32:53+02:00	0cc4m		Vulkan Optimizations and Fixes (#8959)
98a532d474c73d3494a5353024cb6a4fbbabbb35	43bdd3ce188cd247bda7c1bd1ad01fa64e566690	2024-08-14T02:51:02-04:00	compilade		server : fix segfault on long system prompt (#8987)
8cd1bcfd3fc9f2b5cbafd7fb7581b3278acec25f	a21c6fd45032a20180e026773582d21294c85619	2024-08-11T16:58:58+03:00	Georgi Gerganov		flake.lock: Update (#8979)
272e3bd95e620d285b2fb9faaa7b6b1b8edbbf3a	45a55b91aa87958a75d691be64b070266d4fbb94	2024-08-09T18:24:30+03:00	Georgi Gerganov		make : fix llava obj file race (#8946)
be55695eff44784a141a863f273661a6bce63dfc	0478174d5959b66096ae6609fcb0df14cab66b51	2024-08-07T13:29:02+02:00	slaren		ggml-backend : fix async copy from CPU (#8897)
db20f50cf4710c46e3a996919a26858cae8c80ed	efda90c93a62274ec0b0bfa80c4eee4bdb6966d0	2024-08-06T17:21:47+04:00	Jaeden Amero		cmake : Link vulkan-shaders-gen with pthreads (#8835)
064cdc265fb63590c7c8f04a609d36ef200d55a7	5587e57a76630651752031223cc7024cb32cf308	2024-08-05T07:52:55+02:00	0cc4m		vulkan : fix Qantized Mat-Vec Mul on AMD GPUs for ncols < 64 (#8855)
ecf6b7f23e664afd7ff856ec39034240ce438daa	01aae2b4975b57a265ce8194928fd87f2d71027e	2024-08-04T03:55:03-07:00	Brian Cunnie		batched-bench : handle empty `-npl` (#8839)
4b77ea95f56a4c49bc995f08eac62a6416875ccc	76614f352e94d25659306d9e97321f204e5de0d3	2024-08-04T05:53:20+03:00	Georgi Gerganov		flake.lock: Update (#8847)
b7a08fd5e0e7c898c68d1743066ea495202d9608	7a11eb3a260915aee16101808f291a244e2facc7	2024-08-01T12:53:46-04:00	Alex O'Connell		Build: Only include execinfo.h on linux systems that support it (#8783)
ed9d2854c9de4ae1f448334294e61167b04bec2a	398ede5efeb07b9adf9fbda7ea63f630d476a792	2024-07-31T15:51:06-04:00	Clint Herron		Build: Fix potential race condition (#8781)
7c27a19b2eb91bb0f43c7f7aec0386cec2dddc33	140074bb8647df41840d6f32f4409fa8959bcf9f	2024-07-30T23:40:18+09:00	l3utterfly		added android implementation of ggml_print_backtrace_symbols (#8751)
140074bb8647df41840d6f32f4409fa8959bcf9f	6e2b6000e5fe808954a7dcef8225b5b7f2c1b9e9	2024-07-30T15:58:57+03:00	Georgi Gerganov		flake.lock: Update (#8729)
439b3fc75a8deb42899ac47a8f52aae75e0339fe	0832de723695ab400316a6c49b9f712380e3a731	2024-07-29T20:56:12+08:00	R0CKSTAR		cuda : organize vendor-specific headers into vendors directory (#8746)
6eeaeba126ff701f3e8f79f246805b7023709972	4730faca618ff9cee0780580145e3cbe86f24876	2024-07-28T22:32:44+02:00	Johannes Gäßler		cmake: use 1 more thread for non-ggml in CI (#8740)
4c676c85e59ef8f771f3a129e6eb217552139231	e54c35e4fb5777c76316a50671640e6e144c9538	2024-07-28T00:42:05-04:00	compilade		llama : refactor session file management (#8699)
e54c35e4fb5777c76316a50671640e6e144c9538	5e2727fe0321c38d1664d26173c654fa1801dc5f	2024-07-28T07:41:25+08:00	R0CKSTAR		feat: Support Moore Threads GPU  (#8383)
45f2c19cc57286eead7b232ce8028273a817aa4d	22f281aa16f44d8f6ec2c180a0685ff27e04e714	2024-07-21T16:45:10+03:00	Georgi Gerganov		flake.lock: Update (#8610)
22f281aa16f44d8f6ec2c180a0685ff27e04e714	328884f4219c0228673cf1870ac63987fb4f9fd0	2024-07-20T22:09:17-04:00	M-A		examples : Rewrite pydantic_models_to_grammar_examples.py (#8493)
69c487f4ed57bb4d4514a1b7ff12608d5a8e7ef0	07283b1a90e1320aae4762c7e03c879043910252	2024-07-20T22:25:26+02:00	Johannes Gäßler		CUDA: MMQ code deduplication + iquant support (#8495)
87e397d00bdcedd5cbf6dfda06a7b0f302462728	57b1d4f9eb6f2c139b31ea79626d954b261e1051	2024-07-19T17:17:27+02:00	slaren		ggml : fix quant dot product with odd number of blocks (#8549)
7acfd4e8d55082c1b597dfc3ffe04fb5d530c6dc	97bdd26eee11fe109dec00de75690ceef61c03f2	2024-07-15T23:13:10-04:00	compilade		convert_hf : faster lazy safetensors (#8482)
f17f39ff9cb71fa7b0eda0c649bb0ce9b2d8a6b8	9104bc20edf47f74dc49379b7d61d0c6e85a4882	2024-07-15T08:04:56-04:00	M-A		server: update README.md with llama-server --help output [no ci] (#8472)
aaab2419eaa17ab3aa38f4ba49c7eea406999e99	73cf442e7bc9baca7b3e213b261551812f1676c9	2024-07-14T18:54:02+03:00	Georgi Gerganov		flake.lock: Update (#8475)
278d0e18469aacf505be18ce790a63c7cc31be26	dd07a123b79f9bd9e8a4ba0447427b3083e9347a	2024-07-10T20:08:17-04:00	Clint Herron		Initialize default slot sampling parameters from the global context. (#8418)
0f1a39f3439825acf7e3a1663566d410be152170	83321c6958acf20747d288031174cc1bf8816e33	2024-07-10T07:14:51-05:00	Dibakar Gope		ggml : add AArch64 optimized GEMV and GEMM Q4 kernels (#5780)
7fdb6f73e35605c8dbc39e9f19cd9ed84dbc87f2	a130eccef42b75a84da270411cefeed45c153e30	2024-07-09T01:36:38+03:00	Georgi Gerganov		flake.lock: Update (#8342)
470939d483d1c89b7292f78bac1fd27c42c171ce	6f0dbf6ab087bcd286fb78560099ca0458316735	2024-07-08T03:26:53-04:00	Kevin Wang		common : preallocate sampling token data vector (#8363)
cb4d86c4d723af87d3d7e3177e9485f200391384	86e7299ef5dff0f388922dc6fcbce009e99d8005	2024-07-07T11:10:38+02:00	Bjarke Viksøe		server: Retrieve prompt template in /props (#8337)
213701b51a17175d0d326b566efc03f30ec7fbe6	be20e7f49d9e5c6d9e8d9b4871eeba3df7a1639d	2024-07-05T19:01:35+02:00	jaime-m-p		Detokenizer fixes (#8039)
d0a7145ba99ed3a8bc3145aa785b5c86ffe65020	9ef07800622e4c371605f9419864d15667c3558f	2024-07-01T02:09:34+03:00	Georgi Gerganov		flake.lock: Update (#8218)
ab3679112d4c49a215a3d31550a7720b202e9015	97877eb10bd8e7f8023420b5b5300bcbdadd62dc	2024-06-27T18:37:29+03:00	Georgi Gerganov		flake.lock: Update (#8071)
9b31a40c6ddabe552875b811d7127aa039ca9703	c70d117c37cc7876e775d1e2722208a50c52edb3	2024-06-27T01:50:09+02:00	Daniel Bevenius		clip : suppress unused variable warnings (#8105)
e6bf007744eb06336a231ef39cf08146dd16d2ce	84631fe1504de40427dc4b4cdac92fa7ebf65955	2024-06-25T21:07:28+02:00	Daniel Bevenius		llama : return nullptr from llama_grammar_init (#8093)
3791ad219323389106dc3fd80814eb5bbb7b80de	f702a90e245499283d6de0b287701c723cda2a87	2024-06-25T16:57:35+05:30	HanishKVC		SimpleChat v3.1: Boolean chat request options in Settings UI, cache_prompt (#7950)
95f57bb5d5b18ef0beb2702a0d6c06e46804075c	e112b610a1a75cb7fa8351e1a933e2e7a755a5ce	2024-06-24T03:07:59+02:00	slaren		ggml : remove ggml_task_type and GGML_PERF (#8017)
e112b610a1a75cb7fa8351e1a933e2e7a755a5ce	6a2f298bd784403c5c33eebb822217ec5d9b5590	2024-06-24T02:27:57+08:00	Eddie-Wang		llama : add support for BitnetForCausalLM (#7931)
b6b9a8e606da7764bd3649c2ea574979c85abd43	45c0e2e4c1268c2d7c8c45536f15e3c9a731ecdc	2024-06-23T13:14:45+02:00	slaren		fix CI failures (#8066)
ba5899315283eb1df3902363daf79bdc5eefe426	a7854743c5e6216a6178824a603aa9479728ddd5	2024-06-19T23:57:10Z	sasha0552		server : fix smart slot selection (#8020)
9c77ec1d74874ee22bdef8f110e8e8d41389abf2	a04a953cab583f0229e7b4b506346e3e9a85c8d0	2024-06-19T15:04:15+02:00	slaren		ggml : synchronize threads using barriers (#7993)
6a2f0b3474d479bda4ac2ee7cfd5dcdcf0be1f79	21be9cab94e0b5b53cb6edeeebf8c8c799baad03	2024-06-17T16:10:15+02:00	Markus Tavenrath		Implement non-mapped async IO for CUDA on Windows.  (#7896)
bc6c457fa35f6791e9a2bb61108e7d49e8fc98bd	52399254b3bceda279b4ea9111a983e32310166e	2024-06-16T19:16:21+03:00	Georgi Gerganov		flake.lock: Update (#7951)
cddaf028adc738b5a7ecc60809cb78e0ba0f97c1	c8a82194a888f68f259e0d0fa96f3332ac7c5cb6	2024-06-16T14:50:12+03:00	Georgi Gerganov		ggml : fix handling of zero blocks in IQ quants (#7955)
0c7b3595b9e5ad2355818e259f06b0dc3f0065b3	7b2f4a7d193ef2475259bbe7656fcccfab4b1217	2024-06-15T18:53:40+02:00	Xuan Son Nguyen		Add `cvector-generator` example (#7514)
76d66ee0be91e2bec93206e821ee1db8d023cff5	66ef1ceedf983773c8ceb4d925285d41d4e50e2a	2024-06-14T18:41:49+02:00	Johannes Gäßler		CUDA: faster q2_K, q3_K MMQ + int8 tensor cores (#7921)
f578b86b2123d0f92afbaa98a031df4d4464e582	1c641e6aac5c18b964e7b32d9dbbb4bf5301d0d7	2024-06-13T03:11:35+02:00	slaren		move BLAS to a separate backend (#6210)
10ceba354a3b152ff425e9fa97f9caaef99a46b1	e95beeb1fc4621826ddd616776dbdf717366bf5c	2024-06-10T02:04:50+03:00	Georgi Gerganov		flake.lock: Update (#7838)
7a16ce7db2a74a223f0f3b9cee66d4539c5bce8f	da799b41891e34aac86ce4e173f9c4c0afd4fab3	2024-06-08T07:50:31Z	sasha0552		server : smart slot selection using Longest Common Prefix (#7728)
27615f5ab21060d96953c9c1e223051ab2188f57	7027b27d765db95d4ac6b569d976e387a8715881	2024-06-07T05:15:07-07:00	intelmatt		cmake : fix BUILD_SHARED_LIBS=ON build (#7784)
ee459f40f65810a810151b24eba5b8bd174ceffe	f83351f9a62a6262f1fc3d08f320033089cddfb5	2024-06-06T19:19:59+03:00	Georgi Gerganov		server : fix --threads-http arg (#7801)
ad675e1c67a05b16e4e12abe30dbecfc808e7b7e	a143c04375828b1f72eb1a326115791b63e79345	2024-06-06T06:08:52-07:00	Clint Herron		Added support for . (any character) token in grammar engine. (#6467)
9973e81c5ccf4f31b3980f5aa73f5cfea8699860	c90dbe026b456a233f8f0fbe752212e6a0503ca2	2024-06-04T23:40:49-07:00	arch-btw		readme : remove -ins (#7759)
6d1616944d9efd342ed2a4fd318722adfc9febcd	bde7cd3cd949c1a85d3a199498ac98e78039d46f	2024-06-04T10:01:09+03:00	Georgi Gerganov		ggml : prevent builds with -ffinite-math-only (#7726)
a5735e4426b19a3ebd0c653ad8ac01420458ee95	0b832d53ba0ffcc759c8d62ede3772dd62321f8e	2024-06-04T00:14:15+09:00	Masaya, Kato		ggml : use OpenMP as a thread pool (#7606)
6f28a333c1e3fdfdc7b4f9d0367f2b41a9b7e9d4	549279d8049d78620a2b081e26edb654f83c3bbd	2024-06-03T15:49:30+08:00	zhangkaihuo		llama : MiniCPM support tied embeddings (#7664)
1669810d7c2446af8425aa54ff6611bf6f14646c	7c4e5b7eae26581869e782015d9deca947c34997	2024-06-03T00:13:12+03:00	Georgi Gerganov		flake.lock: Update (#7686)
2ac95c9d5678d05e253691fb1f26471675bff5ad	750f60c03e4d3f53fa51910551ce87a3d508d2d7	2024-06-01T21:50:18+05:30	HanishKVC		SimpleChat: Simple histogram/repeatMatching driven garbageTrimming, Settings UI, Streaming mode, OpenAi Compat (Model, Authorization Bearer), Save/Restore session, Auto Settings UI (#7548)
972b555ab935705f3437abd5909a5c46852811f6	3854c9d07f67de7f8cd6d86117bfaef47549b05a	2024-05-30T09:52:39+02:00	Johannes Gäßler		README: explain parallel build [no ci] (#7618)
87bdf2a199acd62e19814d7a4d0500a04a7f09f3	00281b7be32462754618c42ed93f95743af46627	2024-05-29T13:36:39+02:00	slaren		ggml : use atomic_flag for critical section (#7598)
ee3dff6b8e39bb8c1cdea1782a7b95ef0118f970	edc29433fa08b4e5aeb67649a29fc7713af13d04	2024-05-28T17:07:05+02:00	fairydreaming		Add support for DeepseekV2ForCausalLM (#7519)
c41767154eb82aa3fe7568fc816c3402b78eae94	74b239b3d5f067470d7ef5e26e2e059720572e32	2024-05-28T00:41:14-04:00	Nathan Epstein		Markdownish code block fix (#7571)
dff451cfa1f297348751ce6b538670e1ae9a7d5b	d298382ad977ec89c8de7b57459b9d7965d2c272	2024-05-26T18:54:56+03:00	Georgi Gerganov		flake.lock: Update (#7540)
d041d2ceaaf50e058622d92921b3e680ffa4e9e7	27891f6db03de6e3fd5941983838c29bef253352	2024-05-24T18:59:06+03:00	Georgi Gerganov		flake.lock: Update (#7232)
fbca2f27fc7fa9aa4a8ad0357478fdb908472908	0df0aa8e43c3378975269a51f9b876c8692e70da	2024-05-24T14:31:13+02:00	fairydreaming		Add support for ArcticForCausalLM (#7020)
3015851c5ac7334fb544a23a70a284c117b87044	55ac3b7aeaf52f19786ed96e885d89521fc0f6c8	2024-05-23T14:29:26+02:00	Daniel Bevenius		llama : add getters for n_threads/n_threads_batch (#7464)
1e374365d170b7f692fd7753c145e21bc14486c8	197ff91462dd05bb9a3be03578114abf0c355536	2024-05-22T23:23:21+05:30	HanishKVC		SimpleChat: a simple and dumb web front end for testing /chat/completions and /completions end points and try chat (#7350)
3bc10cb485dd7efa4da6c64e73ad0c9e2bfe0821	6bf9b66fa3f263ca2175dcb5f6d0a658581e1dfb	2024-05-20T15:10:03+03:00	Georgi Gerganov		server : fix temperature + disable some tests (#7409)
cb42c294279bc4a0a4e926a7b5a5568049f12fa7	d233b507cd19fcc2d8d8963ecc6a3eb7a33f2ecc	2024-05-18T11:10:47+02:00	Johannes Gäßler		server: correct --threads documentation [no ci] (#7362)
d273c1402b25086fd91aef2467ac13f2e49fa0ea	27b040691cbe45314147c2745e891a38e9c048d4	2024-05-17T15:11:45+03:00	Aarni Koskela		py : convert-hf-to-gguf-update improvements (#7340)
e1b40ac3b94824d761b5e26ea1bc5692706029d9	dc020985b8755dd6aa93a2f002f43c3ede808cce	2024-05-15T12:59:12-05:00	kunnis		ggml : use dynamic thread scheduling for matrix multiplication (#6915)
f308ea705974dff62a1fe5367d776ad9d5109239	c3c88f296a72432edb697ac8026dbf2ec18f2b21	2024-05-13T11:01:07+03:00	Georgi Gerganov		metal : tune soft_max number of threads (whisper/0)
541600201e6480f54ae09e58d16b154d4b4b331d	efc8f767c8c8c749a245dd96ad4e2f37c164b54c	2024-05-14T09:33:42+02:00	slaren		llama : disable pipeline parallelism with nkvo (#7265)
988631335a20d06497f58be0b8ba13adb4323a22	f99e1e456eaf69cc38c1982a2693ce41c0f897ef	2024-05-11T04:13:02-04:00	Steve Grubb		server : free llama_batch on exit (#7212)
bc4bba364fb96d908f2698e908648df5e6f55e02	c12452c7aec8a02264afc00196a13caa591a13ac	2024-05-08T21:55:49+01:00	agray3		Introduction of CUDA Graphs to LLama.cpp (#6766)
b3a995b416e13ae3123a117a743e11d0ede0ca4c	bcdee0daa7c5e8e086b719e5eb4073b00df70e01	2024-05-06T18:36:06+03:00	Georgi Gerganov		flake.lock: Update (#7079)
9c67c2773d4b706cf71d70ecf4aa180b62501960	952d03dbead16e4dbdd1d3458486340673cc2465	2024-04-30T12:16:08+03:00	Georgi Gerganov		ggml : add Flash Attention (#5021)
b8c1476e44cc1f3a1811613f65251cf779067636	5539e6fdd1b97e0c37c54d34df67f334fc344a26	2024-04-29T14:40:14-04:00	Clint Herron		Extending grammar integration tests (#6644)
6e472f58e40cd4acf6023e15c75a2700535c5f0b	4dba7e8114d84241c842b986e008af8b88d1a019	2024-04-28T00:18:27Z	github-actions[bot]		flake.lock: Update
7d641c26ac73956a54b204cabefe85c764823678	5790c8dac1a4f0aa80b4efee3b962d8c04c829e8	2024-04-26T09:26:59+02:00	Pierrick Hymbert		ci: fix concurrency for pull_request_target (#6917)
28103f4832e301a9c84d44ff0df9d75d46ab6c76	c0d1b3e03e27634ac2871761f5033cf9324d472d	2024-04-24T11:08:36+02:00	Johannes Gäßler		Server: fix seed for multiple slots (#6835)
e9b4a1bf68c18beff4e33f23ea62c1245b296915	5cf5e7d490dfdd2e70bface2d35dfd14aa44b4fb	2024-04-21T00:17:47Z	github-actions[bot]		flake.lock: Update
8cc91dc63c0df397d644a581b2cbeea74eb51ae0	dbceec87c0221ec952e69448df6a71f1372a7487	2024-04-16T14:55:30-04:00	Justine Tunney		ggml : add llamafile sgemm (#6414)
f184dd920852d6d372b754f871ee06cfe6f977ad	422c2aff1c9735853c9d8f5162104e41a364adc4	2024-04-14T16:55:30+03:00	Georgi Gerganov		flake.lock: Update (#6669)
b804b1ef77351d2a11be945462c6c251710476cb	8228b66dbc16290c5cbd70e80ab47c068e2569d8	2024-04-11T14:51:07+02:00	Pierrick Hymbert		eval-callback: Example how to use eval callback for debugging (#6576)
5dc9dd7152dedc6046b646855585bd070c91e8c8	e11a8999b5690f810c2c99c14347f0834e68c524	2024-04-09T09:16:13+01:00	Carolinabanana		llama : add Command R Plus support (#6491)
beea6e1b16e783a0886e78dec01002a8c00db24d	87fb5b4234d4b9c56ac94cf7aa229c8fd7defdb0	2024-04-08T20:43:30+08:00	Jan Boon		llama : save and restore kv cache for single seq id (#6341)
b909236c0bf0b6e872af95df9490492ecec310ac	e0717e751e12af13f4eedaae8bbbd608e40d7e54	2024-04-07T21:25:30+03:00	Georgi Gerganov		flake.lock: Update (#6517)
8120efee1d9931b514aeb5a047209d576f23286c	a74401f0e5ebb15fa4d8b6619d1baa6ea9179123	2024-04-04T16:59:04+02:00	Pierrick Hymbert		ci: bench fix concurrency for workflow trigger dispatch with sha1 (#6478)
7a2c92637ae265654a68f62e6a7610b358255d3f	4bcd6b959ca3991084ad1d8464caf2a734e29b1d	2024-04-04T11:57:58+02:00	Pierrick Hymbert		ci: bench: add more ftype, fix triggers and bot comment (#6466)
08a0c0206075556e82aca0feafad530dcc5f1426	52604860f93063ef98863921da697576af1c7665	2024-04-03T15:07:05+02:00	slaren		ggml : mul_mat_id use the same tensor for all the experts (#6387)
f87f7b898651339fe173ddf016ca826163e899d8	33a52448061cfd2ea44da9e6cb30b2ec22e2f6d0	2024-04-01T19:05:57+03:00	Georgi Gerganov		flake.lock: Update (#6402)
b75c38166cf0c66793a32d5c3d6cb69929dd083d	bfe7dafc9cf96b9a09ead347fed9a547930fc631	2024-03-29T08:15:00+01:00	Pedro Cuenca		convert : allow conversion of Mistral HF models (#6144)
6902cb7f2e3479f364ee177118200fb7e4e9fc92	d2d8f389960a106b66313ff1621bdb1aaaaaa285	2024-03-28T16:50:48+08:00	Eric Zhang		server : stop gracefully on SIGTERM (#6348)
557410b8f06380560155ac7fcb8316d71ddc9837	55c1b2a3bbd470e9e2a3a0618b92cf64a885f806	2024-03-26T10:46:41-04:00	compilade		llama : greatly reduce output buffer memory usage (#6122)
55c1b2a3bbd470e9e2a3a0618b92cf64a885f806	e097633f63fdd26d492844f7eff056e4083fd9eb	2024-03-26T15:21:27+01:00	Kawrakow		IQ1_M: 1.75 bpw quantization (#6302)
43139cc528909c3de8c144ed174e09a1f7912a80	2f34b865b62b1d2b5eb8a27885e4de220deeacbd	2024-03-25T17:22:27+02:00	Georgi Gerganov		flake.lock: Update (#6266)
f482bb2e4920e544651fb832f2e0bcb4d2ff69ab	1997577d5e121568ae39f538021733ccd4278c23	2024-03-23T18:07:00+01:00	Pierrick Hymbert		common: llama_load_model_from_url split support  (#6192)
1997577d5e121568ae39f538021733ccd4278c23	476b0251b27fb64c575507024a671e639d675594	2024-03-23T18:00:38+01:00	Pierrick Hymbert		server: docs: `--threads` and `--threads`, `--ubatch-size`, `--log-disable` (#6254)
ee804f6223777019cf921e0d99cc24669313ab98	80bd33bc2c4be352697dc8473339f25e1085d117	2024-03-23T02:15:06+09:00	Minsoo Cheong		ci: apply concurrency limit for github workflows (#6243)
be07a03217376c53b1d95e5f658adbbbb2831555	d0a71233fbf8ade8ef06ad8e6b81d1d7b254895f	2024-03-22T06:41:24+08:00	Jan Boon		server : update readme doc from `slot_id` to `id_slot` (#6213)
2d15886bb092c3b780c676b5cc57ff3337af9c83	d199ca79f279e84ebe27caafe0aa59c461d88969	2024-03-17T06:37:44Z	github-actions[bot]		flake.lock: Update
dc0f6125487dcfbff913360f9d877bc0ccf6aa57	c47cf414efafb8f60596edc7edb5a2d68065e992	2024-03-18T01:12:22+08:00	GainLee		ggml:fix finding transfer queue family index error (#6094)
12247f4c69a173b9482f68aaa174ec37fc909ccf	4e9a7f7f7fb6acbddd1462909c8d696e38edbfcc	2024-03-15T16:41:22-04:00	Andrew Canis		llama : add Command-R support (#6033)
f30ea47a87ed4446ad55adb265755dc9102956a2	d8fd0ccf6ac8b07791ffd1575eed436930854ae3	2024-03-13T18:54:21+01:00	slaren		llama : add pipeline parallelism support (#6017)
8030da7afea2d89f997aeadbd14183d399a017b9	184215e783dfad76aded2c68244c327a5c507df5	2024-03-12T14:27:20+02:00	Georgi Gerganov		ggml : reuse quantum structs across backends (#5943)
05b06210c954491cf0f12034b0a62bd4d69ce78b	83796e62bc9f6caae6228168e359890f51e60fee	2024-03-11T17:49:47+02:00	Georgi Gerganov		llama : more consistent names of count variables (#5994)
caa106d4e05a0ab94225c220b81f9e2cd522339b	3202361c5b1ba15e695b31209567ef42c22c5c32	2024-03-11T10:56:41+01:00	Xuan Son Nguyen		Server: format error to json (#5961)
be858f620508385ad12d0e5e862010e666ca729c	ef3ced26a3817d92890b97b83acaeb018ade02d0	2024-03-11T07:51:49+01:00	Kawrakow		Better 1.5 bit quantization  (#5971)
fa8a809a9119411bc9c3f00026550d0343f4d9b7	bcebd7dbf62fd7b293d5ed089023e4e733269c71	2024-03-10T18:17:47+01:00	Pierrick Hymbert		server: ci: windows build and tests (#5968)
c78541479cf835dd9eb568ccd9a2083198a7203d	621e86b331f8b0e71f79fd82a4ae1cd54c3e4396	2024-03-10T16:43:08+02:00	Georgi Gerganov		nix: update flake.lock (#5969)
d894f352bf433157232dc8dc54eacd50014e898e	098dbaab449f5309a54871ba7e5acef72ae696de	2024-03-09T19:55:54+01:00	slaren		perplexity : support using multiple sequences to allow larger batch sizes (#5946)
c2101a2e909ac7c08976d414e64e96c90ee5fa9e	515f7d0d4fce41c752fc253acf30707c3be2531e	2024-03-08T17:31:00-05:00	compilade		llama : support Mamba Selective State Space Models (#5328)
2002bc96bf2cbf5ab981a17d7e994d817c9801f5	ceca1aef0738b57951cd12c603c3477e75312dec	2024-03-07T11:41:53+02:00	Georgi Gerganov		server : refactor (#5882)
21b08674331e1ea1b599f17c5ca91f0ed173be31	6a87ac3a52668e117d97bcea07b529c93188b303	2024-03-05T16:08:35+08:00	Neo Zhang Jianyu		[SYCL] fix mul_mat fault in CI/unit-test (#5862)
67be2ce1015d070b3b2cd488bcb041eefb61de72	231ae28f078c3148d097b301f2145f1e3e816cc1	2024-03-03T14:26:18+01:00	slaren		cuda : fix data race in soft max (#5853)
8ef969afcec1645d2d9c3ab1fc82263bba968989	fa974646e1a2024fc7dc9e6f27cf1f2f5d4a3763	2024-03-03T08:48:36+01:00	Pierrick Hymbert		server : init http requests thread pool with --parallel if set (#5836)
fa974646e1a2024fc7dc9e6f27cf1f2f5d4a3763	9731134296af3a6839cd682e51d9c2109a871de5	2024-03-03T06:11:31+02:00	Georgi Gerganov		flake.lock: Update (#5842)
bbde6eb2561153aabbdfac5001c690fe00cad639	ef2cd694c4155fbf25bae61c5178c47eb3676dba	2024-03-02T17:00:51+02:00	Kawrakow		ggml : IQ3_S improvements (#5829)
c29af7e2252d288f2ea58a7d437c1cb7c0abf160	38d16b142624bdd7c41d9955752b7f7b59c5e048	2024-03-02T01:00:46+05:30	Sourab Mangrulkar		llama : add StarCoder2 support (#5795)
5cb02b4a012bb16c6c699c0c62c05ffa653eee0f	6ea0f010ff6967034528d9e0b8330b9b0f0b7c13	2024-03-01T10:08:08+01:00	Pierrick Hymbert		server: allow to override threads server pool with --threads-http (#5794)
5f706718566e3a5147916dc381f3b99de0ffad47	a693bea1e6762a17b78b6ddf4611e54136941ea2	2024-02-24T11:27:36-05:00	UEXTM.com		Introduce backend GUIDs (ggml/743)
a693bea1e6762a17b78b6ddf4611e54136941ea2	adcb12a9bad87bc96f2f158c95892b3d04aa7ffb	2024-02-28T09:55:37+01:00	Xuan Son Nguyen		server : hit Ctrl+C twice to exit (#5734)
7c4263d4261d6ee6f0539d53eb9e1b4d120ba8af	cb49e0f8c906e5da49e9f6d64a57742a9a241c6a	2024-02-28T10:37:02+02:00	Kawrakow		ggml : make i-quants work with super-blocks of 64 (CPU,Metal) (#5760)
0becb22ac05b6542bd9d5f2235691aa1d3d4d307	c24a2a6e6005e5d424301525a42ba45a4a362d30	2024-02-27T16:34:24+02:00	Kawrakow		IQ4_XS: a 4.25 bpw quantization (#5747)
c39373398803c669056304090050fe3f44b41bf9	e3965cf35aac00d4e24998c8a3d0093ae1d98bd3	2024-02-25T00:17:11Z	github-actions[bot]		flake.lock: Update
930b1780269a69948d106e2d1b838ab7661f679a	d52d7819b8ced70c642a88a59da8c78208dc58ec	2024-02-25T13:50:32+01:00	Pierrick Hymbert		server: logs - unified format and --log-format option (#5700)
d52d7819b8ced70c642a88a59da8c78208dc58ec	12894088170f62e4cad4f8d6a3043c185b414bab	2024-02-25T13:49:43+01:00	Pierrick Hymbert		server: concurrency fix + monitoring - add /metrics prometheus compatible endpoint (#5708)
9e359a4f47c1b2dceb99e29706c9f7403d32ab5e	4c4cb30736582cacb1a164a9d4bc8e17b1014be7	2024-02-24T19:16:04+01:00	Pierrick Hymbert		server: continue to update other slots on embedding concurrent request (#5699)
4c4cb30736582cacb1a164a9d4bc8e17b1014be7	525213d2f5da1eaf4b922b6b792cb52b2c613368	2024-02-24T16:23:52+02:00	Kawrakow		IQ3_S: a much better alternative to Q3_K (#5676)
525213d2f5da1eaf4b922b6b792cb52b2c613368	fd43d66f46ee3b5345fb8a74a252d86ccd34a409	2024-02-24T12:28:55+01:00	Pierrick Hymbert		server: init functional tests (#5566)
1ecea255ebb70750b52688393f37a63606b90e3f	a00a35cef93e057eace8351a667d14d152a91ebc	2024-02-21T15:47:48+01:00	Pierrick Hymbert		server: health: fix race condition on slots data using tasks queue (#5634)
a14679cc30c785e75d38028bae6ec39c6209ddef	6560bed3f066c876682464762cad90f1e28e3f1b	2024-02-21T11:39:52+02:00	Kawrakow		IQ4_NL: 4-bit non-linear quants with blocks of 32 (#5590)
c0a8c6db371cb3e4379900867b948879f5842201	b9111bd209c7b11b0592450a6ed2e0ca545b2c84	2024-02-20T08:48:19+01:00	Pierrick Hymbert		server : health endpoint configurable failure on no slot (#5594)
f0d1fafc029a056cd765bdae58dcaa12312e9879	a0c2dad9d43456c677e205c6240a5f8afb0121ac	2024-02-18T23:38:32-08:00	bmwl		ggml : android and old glibc NUMA incompatibility bugfixes (#5557)
c145f8a132b2fe1d1e65987faddbd9a40bef7a12	689a091bbe0537ee9abff3e15a1d74f5f3561165	2024-02-18T18:39:57+01:00	Pierrick Hymbert		server : slots monitoring endpoint (#5550)
e75c6279d1c8e7abb82a331f5de7124eed402de2	36376abe05a12a8cb3af548a4af9b8d0e2e69597	2024-02-18T17:31:28+01:00	Pierrick Hymbert		server : enhanced health endpoint (#5548)
66c1968f7a2e895675425e875b6589f1233a1b52	1dcc3fde004787e6fc4d84c9de0bb34cd2901a3e	2024-02-18T08:23:16-08:00	Daniel Hiltgen		server : graceful server shutdown (#5244)
c8e0d7efeb7634ecc2e9832e879ab9fca4510e71	8f1be0d42f23016cb6819dbae01126699c4bd9bc	2024-02-18T00:17:07Z	github-actions[bot]		flake.lock: Update
f486f6e1e5e9d01603d9325ab3e05f1edb362a95	60ed04cf82dc91ade725dd7ad53f0ee81f76eccf	2024-02-16T01:31:07-08:00	bmwl		ggml : add numa options (#5377)
0d4177126b0556e202efb85bf3f768be81076400	7930a8a6e89a04c77c51e3ae5dc1cd8e845b6b8f	2024-02-15T09:01:57+01:00	Elbios		llava : fix memory management bug (#5491)
cf45252a7cfcb998bade46a886e20477cecc538a	03bf161eb6dea6400ee49c6dc6b69bdcfa9fd3fc	2024-02-13T15:14:22+02:00	Georgi Gerganov		tests : multi-thread the tokenizer tests (#5474)
97a336507ed9b971d72262bec7e2b8b7016a054a	c88c74f967028ae3d5ebade40ae586d20a961abc	2024-02-11T00:17:31Z	github-actions[bot]		flake.lock: Update
f026f8120f97090d34a52b3dc023c82e0ede3f7d	cd9aea63b577a83def84dbd6dcd90a6fa02af745	2024-02-10T02:53:28-08:00	Ian Bull		metal : use autoreleasepool to avoid memory leaks (#5437)
4b7b38bef5addbd31f453871d79647fbae6bec8a	e00d2a62dd1441e3b089570ec06d05c18800d368	2024-02-10T05:30:19+11:00	Neuman Vong		vulkan: Set limit for task concurrency (#5427)
e5ca3937c685d6e012ac4db40555d6ec100ff03c	e4124c24775f2cb5b3d7acc93bf9dc5471c172ef	2024-02-09T10:48:06Z	Paul Tsochantaris		llama : do not cap thread count when MoE on CPU (#5419)
f3e2b4fa3f81a410ecb7dec929c259ef8d8dbb7d	f68664ac241a6b5c233d8f1051eef20929b06008	2024-02-07T01:15:19-05:00	Justin Parker		server : update `/props` with "total_slots" value (#5373)
8a79c591de9b7ff3242a94f68b7fb5a17ed8c2be	31e790322133a4b1d0684527ea446e765e8a96cf	2024-02-06T04:20:59-05:00	Justin Parker		server : include total "num_slots" in props endpoint (#5349)
4ffc7a17d4e80c5f3f905139cb570ed9b6934fcb	906cff55c2848fda091d888a1585915ec0c9ea9e	2024-02-06T08:16:23Z	Niall Coates		server : various fixes for the prompt field in /completion (#5300)
6fdfa2ecc684000a25a4ad91823bc82a6652b645	a2d60c9158435ae9a6f14632f07f1acf7a3becef	2024-02-05T10:46:06+02:00	Kawrakow		iq2_xxs: tune quantization (#5320)
9392ebd49ea5ae236a55b47cbf6a13247e8a3b8c	5ed26e1fc9fab4ce96ecf2d84183fe45bdcab0d4	2024-02-04T00:17:24Z	github-actions[bot]		flake.lock: Update
e920ed393d989ed35625ddaf182ebb52cda07fcd	52bb63c7082c859c3f1dfc527227e6a95b299c7c	2024-02-03T18:15:00+01:00	0cc4m		Vulkan Intel Fixes, Optimizations and Debugging Flags (#5301)
15606309a05ccf7fadbaad5538cb7c32acb1e06b	b2b9f025e7821e78bd501d75d01838c26de07a57	2024-01-31T21:10:15+08:00	JidongZhang-THU		llava : add MobileVLM support (#5132)
dabcc5b471348e4ae03ddacc41e19ad75fb2f041	f8e9140cb46eebaa867e1184a9946e4840eec772	2024-01-31T13:43:03+01:00	slaren		ggml : limit n_threads to the max n_tasks (#5238)
2aed77eb06a329f0d82bb1c467f4244904d4073f	c82d18e863fcde91b4b1109b1d0c73ea4470c405	2024-01-29T15:45:41+01:00	divinity76		fix typo "RLIMIT_MLOCK" (#5175)
2307523d322af762ae06648b29ec5a9eb1c73032	0f648573dde61c510560f68244f70ece7e60d8c1	2024-01-28T18:03:59+01:00	0cc4m		ggml : add Vulkan backend (#2059)
0f648573dde61c510560f68244f70ece7e60d8c1	b764b8f1d079ba44d912801ce6d29bd0d94d51cf	2024-01-28T21:26:23+05:30	Abhilash Majumder		ggml : add unified SYCL backend for Intel GPUs (#2690)
b764b8f1d079ba44d912801ce6d29bd0d94d51cf	9241c3a2ace544aef708334e54bbdddb90208ee8	2024-01-28T16:54:54+02:00	Georgi Gerganov		flake.lock: Update (#5162)
b2b2bf988c098851b4f3831f0cf38394bff75121	af4980bfedfd8df43b9e4cd1442895e85fee37bc	2024-01-28T09:35:14+01:00	Johannes Gäßler		Tests for min_p, sampling queue (#5147)
7032f4f6349c17a8352f9f93f7d2122f45469e59	5f1925a8cef81eb9b372faaae34b0dd76d5361d4	2024-01-26T11:17:59-06:00	snadampal		ggml : update softmax n_task calculation (#5126)
48c857aa10aea73210a4a72da3f1a6f99269e75d	413e7b0559f922bd4de5e9eec548829d111651b1	2024-01-26T13:42:20+01:00	Xuan Son Nguyen		server : refactored the task processing logic (#5065)
5eaf9964fc797d4585c214db32a463d557f3ed33	d292f4f2047963f558dd516f1baaa71793e9acf2	2024-01-26T05:06:22+09:00	l3utterfly		llama : dynamic temperature sampling (#4972)
bf63d695b804b1c995c7ae4427a8a86936ea6d25	1387ea21178f9f154944013d4dd9764b54c69deb	2024-01-22T03:17:05-07:00	Michael Hueschen		nix: add cc to devShell LD_LIBRARY_PATH
780e24a22eb595b705cbe8284771e9ceff1c4dd2	3ce7e8f8e7ccfce07e5947ac5f1f3f4628cf68ea	2024-01-22T21:15:08+08:00	Reinforce-II		ggml : parallelize FP32 conversion when using BLAS (#5045)
f7276f7500f7ea588836dd1fc6f126334c517878	15bceec2d73d4166340b46b27677c45ac1b4cdad	2024-01-13T17:10:19Z	Someone Serge		workflows: nix-ci: rebuild on flake.lock updates
7dcbe39d36b76389f6c5cd3b151928472b7e22ff	726c0fa9a2da976e9c5d5c51e185d9dd453fc9e5	2024-01-21T14:42:44+02:00	Kawrakow		Add ability to evauate multiple choice tasks  (#5047)
942c0107a7301434c0a5e7da46bc4cf2393aa556	b43ebde3b0ccbc42d9dd782b32e2fd8eb35b43b5	2024-01-21T05:17:27+02:00	Georgi Gerganov		flake.lock: Update (#5054)
7051aacfac0057fa5fac9ea46c55bffc3892d810	2b3b999cacc7ad1207c32fbdf3479a19c06e1a34	2024-01-19T11:39:11+02:00	Kawrakow		winogrande: evaluate log-probs in parallel (#5036)
821f0a271e7c9ee737945245dd7abfa22cc9b5b0	96d7f56d2918ffde1995dbb32392571deb76d7fc	2024-01-18T21:33:05+01:00	Xuan Son Nguyen		server : defer tasks when "slot unavailable" (#5018)
96d7f56d2918ffde1995dbb32392571deb76d7fc	2d5419d08ab1131623e6a1d554607b7663435e87	2024-01-18T21:12:15+01:00	slaren		llama : fix mlock with no-mmap with Metal (#5025)
3e945cc1e9c06d2001031360e4e303e9548fb02c	ad19812cda4062c9f154ef16315df41fbe6a770a	2024-01-18T19:18:21+02:00	Kawrakow		HellaSwag: speed up by parallelizing log-prob evaluation (#5020)
ad19812cda4062c9f154ef16315df41fbe6a770a	682986a08eb5cb04865d2e713449f17304d266d8	2024-01-18T15:33:01+02:00	Georgi Gerganov		perplexity : faster HellaSwag via batching (#5017)
4feb4b33eeb1756e46084a4db9230b279af1a480	959ef0c0df725c013c7f712eaa7790b8e38a8e20	2024-01-16T18:41:42+01:00	Maximilian Winter		examples : add complete parallel function calling example (#4974)
c37b3474e61d609d43cccc3bde5d559e80e4f5d1	158f8c9e21302114bac3c646f80ea85b52ffa0bd	2024-01-16T19:13:54+02:00	Georgi Gerganov		flake.lock: update flake-parts, flake-parts/nixpkgs-lib, and nixpkgs (#4920)
158f8c9e21302114bac3c646f80ea85b52ffa0bd	862f5e41ab1fdf12d6f59455aad3f5dd8258f805	2024-01-16T17:05:19Z	Paul Tsochantaris		metal : localized logic in `ggml_metal_graph_compute` (#4924)
3a48d558a69c88ac17efcaa5900cd9eb19596ac4	7c8d3abd1a17c28fc56b1a4814bc4b29f91d7454	2024-01-16T14:41:27+01:00	Alex Azarov		metal : replace loop of dispatch_async with dispatch_apply (#4934)
e0324285a569d0583cf2f4a07a2402221ee25f58	3e5ca7931c68152e4ec18d126e9c832dd84914c8	2024-01-16T12:04:32+01:00	stduhpf		speculative : threading options (#4959)
bb0c1392479398f9aba86d9ec98db0b95ede6e6d	9408cfdad6b1c090a7e1419d4434edc260b7e47e	2024-01-14T13:26:53+02:00	Georgi Gerganov		llama : check LLAMA_TRACE env for extra logging (#4929)
df845cc982e7e2ea7b9900e29d55b15338faa78d	6b48ed089377330cdb362970a51c1c89b6d857a8	2024-01-13T17:29:43+01:00	David Friehs		llama : minimize size used for state save/load (#4820)
356327feb3f66980ab687040495d722696d98970	ee8243adaa9a9f51ff449213383874e49efe368f	2024-01-13T09:20:46-05:00	Ziad Ben Hadj-Alouane		server : fix deadlock that occurs in multi-prompt scenarios (#4905)
e7e4df031b9e29d4b55a4e0b0295187f6b213db1	584d674be622fbf1578694ada6e62eebedbfd377	2024-01-12T20:07:38+01:00	slaren		llama : ggml-backend integration (#4766)
eab67950068e4b125007d027232c47d2a5831cd0	d8d90aa343c22fe01429d3540e47ded87e9dcb9d	2024-01-11T12:41:39-05:00	Behnam M		server : add `LOG_INFO` when model is successfully loaded (#4881)
7a9f75c38b5e62fe27b8a5a3ed823b4a3714024b	5c1980d8d4c4e0c0af77359f81cc44d90b3f250b	2024-01-11T02:12:05-05:00	Behnam M		server : update readme to document the new `/health` endpoint (#4866)
cd108e641dbdedd8c5641c4cec1762f751f38136	57d016ba2d46a6e22517a31a75cebb48f9e234b6	2024-01-10T14:56:05-05:00	Behnam M		server : add a `/health` endpoint (#4860)
f2eb19bd8bc9f5730d6e05d7a52a9e19bf5ac099	f3f62f0d835d559e80714bbeb05d03125574e3dd	2024-01-03T03:43:19-05:00	Justin Parker		server : throw an error when `slot unavailable` (#4741)
5d7002d4372ebf107cfaf46fcd90df27b204f330	26f3071d714f0b27ad7f021a46a66a1085480258	2024-01-02T04:38:15-06:00	minarchist		server : add --override-kv parameter (#4710)
edd1ab7bc34c10a780ee7f9a4499f7689cdad36d	198ed7ebfc89b8f2b35a8b1655d57bfb57530c1a	2023-12-31T17:42:22Z	Someone Serge		flake.lock: update
68eccbdc5b56f2a2450f9a8463f9934388cafabf	97bbca6e8522d18041fcde6c3d0907a52ce36446	2023-12-29T06:42:26-08:00	Philip Taron		flake.nix : rewrite (#4605)
441f51dca004debf8b275f1bdc08e0f1af7fd8f8	38b3de4658292582a8941a2be5c77b40ce6ac0f2	2023-12-29T19:23:27+09:00	Tamotsu Takahashi		ci : build with CLBlast + ggml-opencl use GGML_API (whisper/1576)
dc68f0054cd279cddddb0cae0c9ef4f9cbaa512a	de8e496437c59e7d1cc84109e3e49a3478aee25a	2023-12-26T21:23:59+01:00	slaren		cuda : fix vmm pool with multi GPU (#4620)
d232aca5a73b290e218a2e48b91023d5e994203f	31f27758faf4a4bd08101a57c7ec3a473f771f86	2023-12-21T21:07:46+01:00	slaren		llama : initial ggml-backend integration (#4520)
880e352277fc017df4d5794f0c21c44e1eae2b84	66f35a2f48e1965a13835a523e677223dbf148be	2023-12-21T18:07:34+01:00	howlger		py : open merges file as 'utf-8' (#4566)
799a1cb13b0b1b560ab0ceff485caed68faa8f1f	fecac45658a99eddc4d6e36ba0310ca8f87a77f0	2023-12-13T13:04:25+01:00	slaren		llama : add Mixtral support (#4406)
da5eaef1f34d0a1f584cd4a092e7691ea46a9d91	5f6e0c0dff1e7a89331e6b25eca9a9fd71324069	2023-12-06T09:08:17+01:00	stduhpf		speculative : support `--color` (#4343)
52c8bc3cf312e1caf02d37bfb9d9d865cbe33594	e4b76bbe316ee50fb17d9ac29e654c0edf830eba	2023-12-05T15:05:51+05:00	MaggotHATE		sampling : custom samplers order (#4285)
23b5e12eb5a76489b4c3ee22213a081da68b1809	d208995c6da66f252d4054c1c5a90eb8ccb7a2f7	2023-12-04T17:04:21+01:00	Daniel Bevenius		simple : update error message for KV cache check (#4324)
880f57973b8e0091d0f9f50eb5ab4cd4e31582ca	8d6d9f033b8101f929e445cf45b39e1557ca7934	2023-12-01T18:42:11+02:00	Georgi Gerganov		llama : fix integer overflow during quantization (#4284)
8d6d9f033b8101f929e445cf45b39e1557ca7934	ef47ec18da469423c276b683dd9b5741cee7023e	2023-12-01T10:41:56+01:00	Daniel Bevenius		py : add requirements file for convert-hf-to-gguf.py (#4277)
ef47ec18da469423c276b683dd9b5741cee7023e	1d144112c0fbbb4ecc07dbcf4f05a380148bd6de	2023-12-01T10:51:24+02:00	Georgi Gerganov		ggml : add ggml_soft_max_ext (#4256)
f43f09366dfd018e4568e23a232aaa8c4f7cfc78	d2809a3ba2780e00fce5a6149a7eda09f1c0e906	2023-11-30T17:25:04-05:00	Ziad Ben Hadj-Alouane		server : add single-client multi-prompt support (#4232)
e2bd725f4b39bc5c6234858d158e01248f5ab5bd	1f5cd83275fabb43f2ae92c30033b384a3eb37b4	2023-11-30T20:50:40Z	rhjdvsgsgks		py : fix oai proxy (#3972)
8406b0924bf323f37d536dee8b8165c1f3d9d11d	b38a16dfcff88d547f78f52d1bea31b84a05aff7	2023-11-28T10:32:03+02:00	Georgi Gerganov		ggml : re-enable BLAS for CPU when src0 != F32 + remove redundant full offload checks in llama.cpp (#4240)
9d5949f04b1f8b76184818abbd99938c2020803f	ff8238f71d56245f4a6dbea693f4ebd81263464d	2023-11-23T12:34:20+01:00	Daniel Bevenius		examples : fix typo in parallel example doc comment (#4181)
f23c0359a32871947169a044eb1dc4dbffd0f405	40a34fe8d034bd484efd79ccbb95059ca6308dcb	2023-11-20T11:35:47+01:00	Galunid		ci : add flake8 to github actions (python linting) (#4129)
532dd74e38c29e16ea1cfc4e7eedb4f2fab3f3cd	e86fc56f7521ca4b18d1d9939e82abd40c2f1c01	2023-11-11T22:04:58-08:00	Richard Kiss		Fix some documentation typos/grammar mistakes (#4032)
48ade94538fa509465d71023e49d07aab0ec8cd5	f28af0d81aa1010afa5de74cf627dcb04bea3157	2023-11-05T08:12:13+01:00	slaren		cuda : revert CUDA pool stuff (#3944)
d6069051de7165a4e06662c89257f5d2905bb156	4ff1046d75e64f0e556d8dcd930ea25c23eb8b18	2023-11-02T18:10:39+01:00	Oleksii Maryshchenko		cuda : use CUDA memory pool with async memory allocation/deallocation when available (#3903)
0e40806c1cb3bdf9955ed807ffbe212be85b4c67	a2758d08e44ce3624d233af4d23c6843e2e735b5	2023-11-01T14:42:01-03:00	bandoti		common : allow caller to handle help/argument exceptions (#3715)
ff3bad83e29e3009010cbc923bebd769055eaa7f	82a6646e0221216c41edcdf99f5a44bb051391f5	2023-10-28T16:41:07+02:00	Erik Scholz		flake : update flake.lock for newer transformers version + provide extra dev shell (#3797)
34b2a5e1ee4fe6295fb4420eb91131d743694c65	6961c4bd0b5176e10ab03b35394f1e9eab761792	2023-10-26T22:53:37+03:00	Georgi Gerganov		server : do not release slot on image input (#3798)
ad939626577cd25b462e8026cc543efb71528472	1717521cdb976a2219888b0e5cba36e210eee9df	2023-10-24T16:10:43-04:00	cebtenzzre		server : add parameter -tb N, --threads-batch N (#3584) (#3768)
1717521cdb976a2219888b0e5cba36e210eee9df	b2f7e04bd312eaf97eee0523aa09d950d585626b	2023-10-24T23:08:20+03:00	Georgi Gerganov		server : do not block system prompt update (#3767)
438c2ca83045a00ef244093d27e9ed41a8cb4ea9	9e70cc03229df19ca2d28ce23cc817198f897278	2023-10-22T22:53:08+03:00	Georgi Gerganov		server : parallel decoding and multimodal (#3677)
0e89203b517c95ec6675eda75d200a60d1e8921d	c67fe68e417f766970fb1feaf2e66458aa24116a	2023-10-18T16:21:57+03:00	Georgi Gerganov		speculative : add tree-based sampling example (#3624)
c67fe68e417f766970fb1feaf2e66458aa24116a	1117d06607d2d885640ac501f05f0aae5494e2c5	2023-10-18T07:21:48-05:00	Jhen-Jie Hong		metal : implement q5_0 and q5_1 kernels (#3648)
2a4bcbacead886996f175f33479d1d874a3e577f	424b6381c4daeed62e6600e0402e72f39845b58d	2023-10-13T12:33:16+02:00	Daniel Bevenius		llama : remove n_threads from llama_decode_internal (#3614)
a8bdd65525ae86dea905e9866ad369b53e30ac14	70c29da118cdb02bfcbd0376c32b5b2236e48e48	2023-10-11T15:42:22-04:00	Michael Coppola		server : add parameter -tb N, --threads-batch N (#3584)
a1202a31ed8c35705bd09fe91c3e7410c619bd70	94e502dfb79430870b42b8e8ee132b4aaa93e4a8	2023-10-08T12:21:19+02:00	Johannes Rudolph		k-quants : fix comments about block sizing (#3499)
9ca79d5cbbc8d43f2bff951404b6a40ff1ee3788	0c731ca4039ccff86ffab90eaae4ca98037c4496	2023-10-06T10:10:13-06:00	Kerfuffle		kv cache slot search improvements (#3493)
a8777ad84e00cda0399e827cdf971e2c3fab1da2	97af49fa395df77e4c18af0e1655b2fee67c9686	2023-10-06T14:16:38+01:00	pudepiedj		parallel : add option to load external prompt file (#3416)
e2583cbc29cd7d6d1403f338842c07dfc0467e6c	e8b8d32e8663ffc55a02c9721af3a5190382cbb0	2023-10-05T15:57:03+04:00	shibe2		CLBlast: Fix handling of on-device tensor data
7f1a0fe709ea1a861da2f3759f58a28bf8953c12	16bc66d9479edd5ee12ec734973554d4493c5dfa	2023-09-29T03:51:52+08:00	Qu Zongfu		ggml : release the requested thread pool resource (#3292)
16bc66d9479edd5ee12ec734973554d4493c5dfa	0512d66670de3f650c579519833c085014b0f200	2023-09-28T21:42:38+02:00	slaren		llama.cpp : split llama_context_params into model and context params (#3301)
0512d66670de3f650c579519833c085014b0f200	0e76a8992c8200237bbc6471a53fb8796b3872f7	2023-09-28T19:31:04Z	Eve		ci : multithreaded builds (#3311)
0e76a8992c8200237bbc6471a53fb8796b3872f7	2db94d98eda56982d80238840b0652b4137a2a84	2023-09-28T20:40:11+02:00	xaedes		train : finetune LORA (#2632)
ec893798b7a2a803466cc8f063051499ec3d96f7	45855b3f1c7bdd0320aa632334d0b3e8965c26c4	2023-09-28T19:04:36+03:00	Georgi Gerganov		llama : custom attention mask + parallel decoding + no context swaps (#3228)
8185710a80531e9ee0c0cb99d3a9c9af1019ab67	7eb41179edc56083ef4eb2df7967ac9ff38b34fb	2023-09-21T10:43:53+02:00	Johannes Gäßler		CUDA: use only 1 thread if fully offloaded (#2915)
a51b68765799e75e17c7622f376bfbeb66f1bd70	76164fe2e65c058e9ee2c3afd0ad6b182ca57e25	2023-09-15T11:09:24+03:00	Georgi Gerganov		metal : relax conditions on fast matrix multiplication kernel (#3168)
4c8643dd6ea1a163bc5979cb69c1e7ab0975bc93	35f73049af6c676a106a5a990a819ae0bc3fcd7d	2023-09-15T00:32:10+08:00	jameswu2014		feature : support Baichuan serial models (#3009)
cb6c44c5e045709b6bb5cc9bb8c9be107c771a78	a21baeb12202a9020b48c53beaaf4b355228e8ba	2023-09-08T14:09:21+02:00	Przemysław Pawełczyk		build : do not use _GNU_SOURCE gratuitously (#2035)
be8c9c245bd129ebabb80e0a7a8dd7daeb4d30af	be6beeb8d75294552c4918fce06d7b84eebf3d79	2023-09-07T15:45:01+02:00	Kawrakow		metal : parallel RoPE on Metal (#3024)
d59bd97065cd7ded6c4ecab54b1d5e0b1b11e318	35938ee3b0c16f1fbbf240dae21e0228864b938c	2023-09-05T09:55:33+02:00	Kawrakow		Guard against all weights in a super-block being zero (#3010)
35938ee3b0c16f1fbbf240dae21e0228864b938c	921772104ba2219bfdc2b2980d05ebc0aa0c92a4	2023-09-05T10:46:39+03:00	Georgi Gerganov		llama : update logic for number of threads when using BLAS
e8d91589258f9204397a7ac5f9b3c857835c98f8	bce1fef328941499dc0acb76cc7fd7ac90449c2f	2023-09-01T11:15:57+03:00	Kawrakow		metal: somewhat faster f16 x f32 matrix multiply kernel (#2951)
8341a25957b319a03d4a811176cd5ad7f2b0fbd4	849408957c687cde4ab32c147107f643fc55130b	2023-08-30T08:29:32+02:00	staviq		main : log file (#2748)
44c117f41ee01c5ac8fb86bba041f08d8b87b46d	43033b7bb4858da4f591715b3babdf906c9b7cbc	2023-08-28T21:51:47+02:00	xaedes		train : mem usage and other improvements  (#2439)
92b1bbd2ec43c82ec0530ba3c8758846c5790c75	dd0dc366dab10e8df28d3924e7f313b5c695e908	2023-08-28T13:23:55+02:00	Johannes Gäßler		CUDA: fix RoPE asserts, block sizes (#2833)
f55538c3ccba9b926846ef862fa830cea08c433e	ebcee207b6058b7f695bb5c203ad87b1066a9790	2023-08-28T10:59:08+03:00	Georgi Gerganov		metal : fix memory leak (#2762)
c10704d01e21e3dbe4d6ca1026ebff85349dd239	230d46c723edf5999752e4cb67fd94edb19ef9c7	2023-08-27T18:55:41+03:00	Georgi Gerganov		llama : fix MPI threads (close #2827)
789c8c945a2814e1487e18e68823d9926e3b1454	c1ac54b77aaba10d029084d152be786102010eb2	2023-08-27T09:03:27+02:00	slaren		ci : add LoRA test to CI (#2650)
730d9c681e339b76407659344e5a2cd50af7d7d5	c7d92e6dfec3f54849f3a0ba373054d29f321ea2	2023-08-26T14:13:36-06:00	Kerfuffle		convert.py : advanced option (#2753)
38b16dfca6e5032e6cfb90c1653bf1ba4cf647b4	8f8c28e89cb9531211783da697d6e7c445e2af1d	2023-08-24T12:27:25-04:00	Shouzheng Liu		metal : bug-fix when enable ggml-alloc (#2757)
cf658adc832badaaa2ca119fe86070e5a830f8f6	a192860cfec89a38d59a943623bf595b1fe4495b	2023-08-23T23:08:04+03:00	Georgi Gerganov		llm : add Falcon support (#2717)
6381d4e110bd0ec02843a60bbeb8b6fc37a9ace9	dadbed99e65252d79f81101a392d0d6497b86caa	2023-08-21T23:07:43+03:00	Georgi Gerganov		gguf : new file format with flexible meta data (beta) (#2398)
a872a2b28eaefc8d464eaa535c94deeb501666f9	0919a0f73d95cfb93a1646a1d1741a0615fe2c5e	2023-08-17T03:35:53-04:00	Shouzheng Liu		ggml-alloc : fix discrepency between measure&eval (#2639)
fc8ef549e50087762a0b4f901cd74b2defcc6ae3	bf83bff6742c0f1795b4c18695a13a34ac7adf62	2023-08-16T16:08:28-04:00	Shouzheng Liu		metal : enable ggml-alloc (#2627)
9ca4abed893685692f90413e4d43153af12342d9	e59fcb2bc129881f4a269fee748fb38bce0a64de	2023-08-10T13:11:36-07:00	DannyDaemonic		Handle `ENABLE_VIRTUAL_TERMINAL_PROCESSING` more gracefully on earlier versions of Windows.
93356bdb7a324a8f6570f99d02af392cd4c45796	60baff7c8584ec369e53469cad5f92e102b1efe4	2023-08-07T14:25:58+03:00	Georgi Gerganov		ggml : mul mat tweaks (#2372)
f6f9896ac3d2ff207e18f87dab85d126ceef5236	34a14b28ff7f3c98730339bacee035091b2a812a	2023-08-07T10:52:57+03:00	Georgi Gerganov		metal : fix out-of-bounds access + inc concurrency nodes (#2416)
5f631c26794b6371fcf2660e8d0c53494a5575f7	415e99fec27be5a2e4283f1937afd17eb33fbd66	2023-08-04T06:37:24-05:00	Stephen Nichols		Fixing race condition in server and partial stream handling in frontend. (#2391)
1a941869cbef8e9cc351a6c6987e4ae3b0f021f7	b5472ea0ada081a6e1c06998ebbc9a24aa2cd4a4	2023-07-27T11:00:54+03:00	Georgi Gerganov		metal : disable graph concurrency optimization due to bug (#2413)
1aa18ef994a6a2b531434eb13251ef48e56d345b	9a08eaf3c4010962d0126e9e5bfbe9af64b2ac90	2023-07-25T08:00:19-04:00	Shouzheng Liu		metal : concurrently dispatch commands (#2358)
57921ca6db53e08eb90010fba99add85be28b5a1	3602ac4255fd4cd589821346290b464a64b955d1	2023-07-23T21:33:02+08:00	wzy		common : n_threads == -1 uses std::thread::hardware_concurrency() (#2347)
e76d630df17e235e6b9ef416c45996765d2e36fb	1d0824b2476e7fda09751a0235c9e571b76d6f2c	2023-07-23T15:09:47+03:00	Georgi Gerganov		llama : grouped-query attention + LLaMAv2 70B support (#2276)
417a85a0010519224cf154eb85d383ffeafeeead	294f424554c1599784ac9962462fc39ace92d8a5	2023-07-20T06:32:22-04:00	Shouzheng Liu		metal: minor q4 optimization and reduce code size (#2248)
32c54116318929c90fd7ae814cf9b5232cd44c36	ff5d58faecf1f02b05bd015bdfc6a394cf2bc9ba	2023-07-13T21:58:25+08:00	Howard Su		Revert "Support using mmap when applying LoRA (#2095)" (#2206)
a7e20edf2266169ccd97a4eb949a593d628fbd64	1d656d6360359cfdaaf5d64ed9690047b600dbcb	2023-07-07T21:23:57+03:00	Georgi Gerganov		ci : switch threads to 1 (#2138)
72421402834141df6cbdcf595fe46dbd11874dce	3e08ae99ceb143d67f9273fda47541e9d98ff23f	2023-07-07T18:36:37+03:00	Georgi Gerganov		ggml : remove sched_yield() call in ggml_graph_compute_thread() (#2134)
7ee76e45afae7f9a7a53e93393accfb5b36684e1	acc111caf93fc6681450924df9f99679c384c59e	2023-07-04T10:05:27-04:00	Tobias Lütke		Simple webchat for server (#1998)
9d23589d638dc74577d5ff880e6d4248b795f12e	0be54f75a6c3e9a09ea71bdfcdabf9a996a0549b	2023-06-27T19:06:33+02:00	Erik Scholz		fix pthreads setaffinity usage on android (#2020)
b853d456018b10820686362af41b2f2f75f1eec6	9225baef71407d799a6f7f563b77fd7f82791416	2023-06-26T13:57:59-04:00	zrm		ggml : add NUMA support (#1556)
6769e944c727c63612dcafbef52009d21ae00fff	cbebf61ca7584e9709265395f0127ae7fc0f1882	2023-06-26T19:43:07+03:00	Kawrakow		k-quants : support for super-block size of 64 (#2001)
2c9380dd2f77e41149340f3ecb09764d793b16db	051e1b0e6a6e3aee7d989b47760980e6fda5861c	2023-06-17T19:15:02+02:00	Johannes Gäßler		Only one CUDA stream per device for async compute (#1898)
794db3e7b982fee37e3995db9c3a216a57ff65e3	5ddf7ea1fb42bac21026de2f77e0f9c069b92234	2023-06-17T07:53:04-04:00	Randall Fitzgerald		Server Example Refactor and Improvements (#1570)
4bfcc855abdb2c9fcc3c5a84747974521909fa41	6b8312e7979b852f6b6ac9d29cd51fda16c17948	2023-06-15T20:29:48+03:00	Georgi Gerganov		metal : parallel command buffer encoding (#1860)
e32089b2c20b1b87b22912f4a8b93fe01647d5b9	2347e45e7bdb09c9a7d74b2c0bc86c2b65f0c343	2023-06-13T21:04:40+02:00	xaedes		train : improved training-from-scratch example (#1652)
74a6d922f12ccfe16b0c265f43be8978c6f25e98	e4caa8da59c1c97dc23fa336f4d726984a20560f	2023-06-12T22:39:21+03:00	Kawrakow		Metal implementation for all k_quants (#1807)
4f0154b0bad775ac4651bf73b5c216eb43c45cdc	ef3171d16241c18581d4d08374f0b9e396ade6b7	2023-06-10T01:59:17-06:00	Kerfuffle		llama : support requantizing models instead of only allowing quantization from 16/32bit (#1691)
245fc3c37da5ac5963f9f11a9f4f2ac08d96afc6	72ff5282bf0388c60821f504c4c8cc2b1f491aa6	2023-06-09T10:39:59+03:00	Kawrakow		metal : faster q4_0 (#1775)
17366df842e358768c0df7024484fffecfc7865b	44f906e8537fcec965e312d621c80556d6aa9bec	2023-06-06T21:33:23+02:00	Johannes Gäßler		Multi GPU support, CUDA refactor, CUDA scratch buffer (#1703)
d5b111f53d14972669eb52055f9df2567663ad8b	2d43387dafe9c60f15f57aa23ee0b37864b98b32	2023-06-07T01:00:01+08:00	LostRuins		Clblast fixes + enhancements to save VRAM and offload more layers (#1675)
ecb217db4fcfa3880300ad08531a5fb6bb142d45	dcb2ed48268e421baf25adc00d602dad0f415564	2023-06-04T23:34:30+03:00	Georgi Gerganov		llama : Metal inference (#1642)
dcb2ed48268e421baf25adc00d602dad0f415564	d8bd0013e8768aaa3dc9cfc1ff01499419d5348e	2023-06-04T08:12:05+02:00	0cc4m		OpenCL: Fix duplication of layers in VRAM and RAM, add GPU mul kernel (#1653)
1fcdcc28b119a6608774d52de905931bd5f8a43d	ac7876ac20124a15a44fd6317721ff1aa2538806	2023-05-25T23:07:29+02:00	Johannes Gäßler		cuda : performance optimizations (#1530)
7d873811f31d4d8c909015c946a862c0089cda7d	2e6cd4b02549e343bef3768e6b946f999c82e823	2023-05-23T18:01:15+02:00	Maarten ter Huurne		Fix handling of "invalid property" when creating OpenCL command queue (#1565)
79b2d5b69d80be0bf29312fb9a95854876b0a8a5	13c351ad7292c5b5ab35db25c7a4f993e75d9cfd	2023-05-14T17:55:02+02:00	xaedes		ggml : alternative fix for race condition bug in non-inplace ggml_compute_forward_diag_mask_f32 (#1454)
13c351ad7292c5b5ab35db25c7a4f993e75d9cfd	60f8c361ca26328ef8523dfb08077fe2f1034490	2023-05-14T18:22:50+03:00	Georgi Gerganov		ggml : various fixes (#1450)
66841fdb0eaf0cc210757cc7f683d0f4eebadc21	905d87b70aa189623d500a28602d7a3a755a4769	2023-05-13T16:48:03+03:00	Georgi Gerganov		ggml : multi-thread mul and diag_mask ops (#1428)
905d87b70aa189623d500a28602d7a3a755a4769	f954edda935a70a14cf0cc45ecc7fe7d60cf3e4b	2023-05-13T15:38:36+02:00	Johannes Gäßler		ggml : GPU-accelerated token generation (#1412)
f954edda935a70a14cf0cc45ecc7fe7d60cf3e4b	f048af0230ad5381bb20b9e3c1467ec6fc7debdf	2023-05-13T14:56:40+02:00	xaedes		ggml : implement backward pass for llama + small training-llama-from-scratch example (#1360)
b9fd7eee57df101d4a3e3eabc9fd6c2cb13c9ca1	b608b55a3ea8e4760c617418538465449175bdb8	2023-05-12T00:23:08+03:00	Georgi Gerganov		ggml : remove bit shuffling (#1405)
58b367c2d757c0ea12aec672382462b42204c724	ea3a0ad6b6b5ca4693b94acd4cb32e2803f66fae	2023-05-01T18:11:07+02:00	slaren		cuBLAS: refactor and optimize f16 mat mul performance (#1259)
ea3a0ad6b6b5ca4693b94acd4cb32e2803f66fae	2bdc09646d8c6cb74a6f573e9081586b4b83b9d1	2023-05-01T08:58:51-04:00	xloem		llama : update stubs for systems without mmap and mlock (#1266)
a5d30b1f53677cb50791fec41c43e93274347303	76a884920aa1d2fc0dc7a7ac12dfc5ec5816377c	2023-04-30T14:41:35-04:00	jon-chuang		common : better default number of threads (#934)
7296c961d9303010a2b98379f738da2a8a55aa1b	78ec543733d10a1629f984fd0302fdaa4e87fe66	2023-04-28T16:57:16+02:00	0cc4m		ggml : add CLBlast support (#1164)
7a32fcb3b29f4db8aed8a85dc58eb958fb118153	dd0eabc049fb1efc631cab8eb0a646808d704e18	2023-04-25T23:40:51+03:00	Georgi Gerganov		ggml : add Q8_0 quantization format (rename the old one to Q8_1) (ARM NEON) (#1179)
50cb666b8a2e35a49b08c0f6bc81138c8f6f2ac1	25d7abbd1f73582b7e0fdc422a936e8541c0780b	2023-04-21T21:59:17+02:00	slaren		Improve cuBLAS performance by using a memory pool (#1094)
018f2279f5fe3ef743bd8254b23ea8f0efae7e73	9411288271ab548216902a029f42a0a38ebcedb7	2023-04-22T02:27:06+08:00	源文雨		cmake : link threads publicly to ggml (#1042)
1bfc153e2f35ddd9d64b084e8d1a5e6fa57ad1c9	3d59769c3bb7e72c915646ddb1e239b1face19f5	2023-04-21T17:18:26+02:00	Kawrakow		ggml : a faster version for Q4_1 x Q8_0 dot products (#1083)
38de86a7114c97ecf3644e3a60159f1ed893e1b0	e0305ead3a072db9c08b35c9600c49273b38a4b5	2023-04-20T19:42:27+02:00	Kawrakow		llama : multi-threaded quantization (#1075)
f7d05095b404b5500b4a702ea16f67fc22446e49	884e7d7a2bfd7325b107442d6758983f5886ed3d	2023-04-19T20:20:14+02:00	Kawrakow		Q4_2 quantization with rmse-optimized scale and quants (#1062)
66674012389f9537140044290f8517bc4a5e0b74	77a73403ca8eaced2590559d0f9cebd2b3649d32	2023-04-19T00:53:24+02:00	slaren		Multi-threaded ggml_cpy (#1035)
77a73403ca8eaced2590559d0f9cebd2b3649d32	50a8a2af97cb92e53e7a3195aa201c3d87da5415	2023-04-18T23:54:57+03:00	Georgi Gerganov		ggml : add new Q4_2 quantization (ARM only) (#1046)
723dac55fa2ba7adc6e3fc8609781d1ad0378906	0f07cacb05f49704d35a39aa27cfd4b419eb6f8d	2023-04-14T00:03:03-07:00	comex		py : new conversion script (#545)
a3a2a0eda8828b60436e9f69d9ac2c1060d03e7a	d990e3fffc5b0f5448e90a16c79a4f2675100af0	2023-04-13T18:36:40+03:00	Georgi Gerganov		ggml : add GGML_DEFAULT_N_THREADS
95ea26f6e92d620a5437f576b80868aee7f808d6	82d146df9b43cf677e0dbce20b03cf864958a0cc	2023-04-13T14:46:23+02:00	SebastianApel		benchmark : add tool for timing q4_0 matrix multiplication (#653)
2663d2c6784ad7b77998c6874df25648d597f74b	a0caa34b162449b5c13b8d604573053300ff54a1	2023-04-11T06:19:54-07:00	comex		Windows fixes (#890)
f963b63afa0e057cfb9eba4d88407c6a0850a0d8	aaf3b23debc1fe1a06733c8c6468fb84233cc44f	2023-04-08T12:24:37-07:00	comex		Rewrite loading code to try to satisfy everyone:
eeaa7b0492fc79baab8bb1fe195d6c87159f2bd3	986b6ce9f99503c51ec5afd8a10baa32359434c6	2023-04-05T22:11:03+03:00	Georgi Gerganov		ggml : multi-thread ggml_rope() (~3-4 times faster on M1) (#781)
6f23ba5ee235cbcb1eedd63b98422dd8d4392a78	78ca9838ee36660a776e97e3391b6fb5dcaacf7f	2023-03-30T01:53:36-07:00	Justine Tunney		Ensure --mlock works properly with mmap() support
78ca9838ee36660a776e97e3391b6fb5dcaacf7f	a017390358cdb23fffb30988dc84bb190d0403ca	2023-03-29T13:51:37-07:00	Justine Tunney		Make loading weights 10-100x faster
41318d708ed196ff727dce14d263a64b23c7333d	a6956b25a1c783e5e96fe06c9c00438f846ef047	2023-03-29T18:10:07+02:00	Maël Kerbiriou		llama : use the same threshold for OpenBLAS and ggml thread limiting (#577)
c1f885067c61191a07a1aedf684168dda62f3f71	e0670260fb50a882b37074112b1881fb0820cf77	2023-03-28T15:56:03Z	Stephan Walter		ggml : introduce structs for the q4 data blocks (#356)
ecbe466a364876927994e2f1ec14f4d82301d201	502a400192013d3e95ed87b777e8fa3bec45713c	2023-03-25T19:47:21+02:00	Georgi Gerganov		Retire the ggml_mul_mat() branch for transposed src0 (#500)
4640eff23d341a0273587800e17ff4a378132d60	ab77d7631211b299cb734bea6ad1f74324154150	2023-03-25T17:03:10+02:00	Georgi Gerganov		Don't interefe with BLAS for large prompts by running only 1 thread
563cdc391dde140f1084d1012234e8e6f57f881f	8d4a855c241ecb0f3ddc03447fe56002ebf27a37	2023-03-24T08:19:05-07:00	comex		Support calling mlock() on loaded model data on Linux and macOS (#453)
483bab2e3d4a868fe679d8bb32827d2a4df214dc	404e1da38ec8025707031a8027da14dc1590f952	2023-03-23T23:22:01+02:00	Georgi Gerganov		Avoid the transposed X branch in the Z = X * Y matrix multiplication (#439)
ae44e23ee36c02da0e37ab508a4b473ace724f8e	928480ef5b7b03d7a07e98286aebe3d8b24457d9	2023-03-22T07:47:15+02:00	Georgi Gerganov		When seed <= 0 - use the clock to generate one
2e664f1ff413995506c9a54f3a8d5b8c64e37a91	8cf9f34eddc124d4ab28f4d2fe8e99d574510bde	2023-03-21T07:35:42-07:00	Casey Primozic		Add initial AVX512 support for dot product on Linux (#320)
7392f1cd2cef4dfed41f4db7c4160ab86c0dfcd9	ad5fd5b60cfdfbfb22b0f2bc9e9f6c9692768f8d	2023-03-19T12:38:44-06:00	Suaj Carrot		Improved quantize script (#222)
4f546091102a418ffdc6230f872ac56e5cedb835	e81b9c81c101f64531ef0fa1ee6b77d562635652	2023-03-17T21:46:46+02:00	Georgi Gerganov		Default to 4 threads (#243)
367946c668757532deed929e1d78673c6ac6bcb8	6b0df5ccf360fe5c015f6607f0375bfc6849005e	2023-03-17T17:47:35Z	Stephan Walter		Don't tell users to use a bad number of threads (#243)