823 lines
140 KiB
Plaintext
823 lines
140 KiB
Plaintext
fd9bd632f346085920d523dd307a3e4c11cc0a05 e08e473cf292ba73789769b993bb35bcdfc31689 2026-06-01T14:15:35+08:00 wangbomeng llama-bench: fix device-info
|
||
e08e473cf292ba73789769b993bb35bcdfc31689 f86112d6758298241ab2fa84ad4f0f7b1ace35c6 2026-06-01T11:47:01+08:00 wangbomeng llama-bench:add device-info
|
||
f86112d6758298241ab2fa84ad4f0f7b1ace35c6 d1bc743c10080fed1253686ce01749b42611ded4 2026-05-29T16:40:38+08:00 wangbomeng add dump_pos,fix pos_tensor of pre_by_embeds
|
||
d1bc743c10080fed1253686ce01749b42611ded4 2f201160b1960fd7be18b70d5770599d7082dd2f 2026-05-28T14:23:06+08:00 wangbomeng set u_batch = max_seq_len
|
||
2f201160b1960fd7be18b70d5770599d7082dd2f f131bf1908b8c5dd1f49d7ae7f616506fc471666 2026-05-27T09:58:31+08:00 wangbomeng delete useless code
|
||
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
|
||
33d96dae28e17208ef61a71dba40cda3c04f135a 37f068d87cc2d0a0b40ce978031fc9932fc89077 2026-05-22T14:33:45+08:00 wangbomeng add llama-build.sh
|
||
37f068d87cc2d0a0b40ce978031fc9932fc89077 13508e5e18bbb1202ae60c2ee5ecbdda1cca817b 2026-05-21T14:07:00+08:00 wangbomeng release
|
||
13508e5e18bbb1202ae60c2ee5ecbdda1cca817b b6b7919468126d5ba30a10941825154789e9082f 2026-05-21T11:43:59+08:00 wangbomeng reduce cmake log
|
||
b6b7919468126d5ba30a10941825154789e9082f 46847c24fc5c52dd57bb4fb17fc4d684149bb4fa 2026-05-20T16:40:33+08:00 wangbomeng reduce cmake log
|
||
46847c24fc5c52dd57bb4fb17fc4d684149bb4fa d028722a12c5a463cc97207787cfd03d7a2590b0 2026-05-20T16:02:12+08:00 wangbomeng Simplify CMakeLists.txt
|
||
d028722a12c5a463cc97207787cfd03d7a2590b0 63442fde8dc285817f725bd6c5fae80f478a3813 2026-05-20T14:41:05+08:00 wangbomeng replace cparam.n_ubatch with n_ubatch()
|
||
63442fde8dc285817f725bd6c5fae80f478a3813 fc3f4a9fab88e98eff8eeca8cbc6617333edba2c 2026-05-20T11:47:42+08:00 wangbomeng fix ubatch and cmakelist
|
||
fc3f4a9fab88e98eff8eeca8cbc6617333edba2c c931ef3163940227c9b6291e04216245cce21429 2026-05-19T16:54:16+08:00 wangbomeng add dump and rt case generation
|
||
c931ef3163940227c9b6291e04216245cce21429 a43741244a646d3f8fa3ff01bb9b7d059223d0a5 2026-05-18T17:34:37+08:00 wangbomeng calrt: fix prefill_by_ids.fix create_buf
|
||
a43741244a646d3f8fa3ff01bb9b7d059223d0a5 b2003d6c90ef0fa26ab9e930ba3e22b355eb3d39 2026-05-15T14:39:53+08:00 wangbomeng run llama-passkey successfully, fix llama-bench, copy 3 times input to infer qwen2.5vl
|
||
b2003d6c90ef0fa26ab9e930ba3e22b355eb3d39 7800c772d44a7415640ec1669a691232939927cf 2026-05-13T15:24:24+08:00 wangbomeng try passkey and embedding
|
||
7800c772d44a7415640ec1669a691232939927cf dde8b8228081769c8d8f1e0751d47fa7875f8423 2026-05-13T15:23:54+08:00 wangbomeng try passkey and embedding
|
||
dde8b8228081769c8d8f1e0751d47fa7875f8423 fabcc7dac4b34d83fe430980bb364b81087c7ee9 2026-05-12T10:46:08+08:00 wangbomeng clip:add minicpm patch;
|
||
fabcc7dac4b34d83fe430980bb364b81087c7ee9 1e9787962f20a7e82c71589ac1dd0585454e4bfe 2026-04-30T16:17:17+08:00 wangbomeng calrt: fix encode dump
|
||
1e9787962f20a7e82c71589ac1dd0585454e4bfe f584311c716dd078c4b737eebbfda97fe12b7274 1bec0db5a675ddc60fb793be2a746e8f3c8855dc 2026-04-30T16:06:23+08:00 wangbomeng Merge remote-tracking branch 'origin/dev-ben' into dev
|
||
f584311c716dd078c4b737eebbfda97fe12b7274 a339954cc2a145a80c5ea349e7896211916cbed9 2026-04-30T16:05:15+08:00 wangbomeng calrt: fix encode dump
|
||
1bec0db5a675ddc60fb793be2a746e8f3c8855dc 465df20c3e1f3f58d0f5531c796e0941a49c32b0 2026-04-30T16:02:59+08:00 wangbomeng bench: support 2 calbins
|
||
465df20c3e1f3f58d0f5531c796e0941a49c32b0 a339954cc2a145a80c5ea349e7896211916cbed9 2026-04-30T09:08:12+08:00 wangbomeng support one calbin
|
||
a339954cc2a145a80c5ea349e7896211916cbed9 8e76fc330ad624d1768b412d7894e272dbe696f4 eeeef3f6d6a5aad9296323ca15d9d929745b8932 2026-04-29T14:57:55+08:00 wangbomeng Merge branch 'dev' into dev-cli
|
||
8e76fc330ad624d1768b412d7894e272dbe696f4 431ad6f1c787f1b47e9e7be14b2e1a7d8565485c 2026-04-29T14:56:59+08:00 wangbomeng finish dev of llama-cli
|
||
eeeef3f6d6a5aad9296323ca15d9d929745b8932 431ad6f1c787f1b47e9e7be14b2e1a7d8565485c 2026-04-28T16:16:19+08:00 wangbomeng add unknown time info: prefill and decode
|
||
431ad6f1c787f1b47e9e7be14b2e1a7d8565485c efa1876bb8b2a449a55054a8c3745892f2c0f262 2026-04-27T16:52:15+08:00 wangbomeng update calrt_infer
|
||
efa1876bb8b2a449a55054a8c3745892f2c0f262 4a2a3181f1cea170105c771e6ba69d851b82b937 2026-04-27T11:35:58+08:00 wangbomeng update version print
|
||
4a2a3181f1cea170105c771e6ba69d851b82b937 e3ea5541259f35871c3e995331770aaca297e9ec 2026-04-26T10:56:02+08:00 wangbomeng add calrt-version
|
||
e3ea5541259f35871c3e995331770aaca297e9ec 0b8e43943846a1cd05c26c9186b9ef778ebae6e2 2026-04-26T10:55:48+08:00 wangbomeng add calrt-version
|
||
0b8e43943846a1cd05c26c9186b9ef778ebae6e2 5eb46fffbd2e260c540fb112314ad67bbb8f56b6 2026-04-26T08:43:43+08:00 wangbomeng to debug
|
||
5eb46fffbd2e260c540fb112314ad67bbb8f56b6 005112c51e00d998b457337ca3dad53fc46e2d0b 2026-04-24T16:24:51+08:00 wangbomeng clip: add image_to_patches; calrt: vision copy data from bf16 to fp32
|
||
005112c51e00d998b457337ca3dad53fc46e2d0b 509670642e8090a1713f5d73590b549be827aaef 2026-04-24T16:24:31+08:00 wangbomeng clip: add image_to_patches; calrt: vision copy data from bf16 to fp32
|
||
509670642e8090a1713f5d73590b549be827aaef 5d219bfcef239c1e3ddd1baecc0514a9462ff6a8 2026-04-24T09:40:34+08:00 wangbomeng rm debug code
|
||
5d219bfcef239c1e3ddd1baecc0514a9462ff6a8 6787a53ab9b4897c5a7b51a4062d7b90e6697d35 2026-04-24T09:31:32+08:00 wangbomeng calrt: recover base = batch_index * dict_len
|
||
6787a53ab9b4897c5a7b51a4062d7b90e6697d35 e1a76abf66178ea85656479a5e19aea7785c09f4 2026-04-24T09:22:02+08:00 wangbomeng calrt:input of multimodel from fp32 to bf16
|
||
e1a76abf66178ea85656479a5e19aea7785c09f4 99dd308adb8488fa869bb669e3335e646a938eff 2026-04-22T16:29:22+08:00 wangbomeng calrt:add CalcoreRT version print
|
||
99dd308adb8488fa869bb669e3335e646a938eff 06c7852e99e34c71e24e3ef6001cb54c72970e4d 2026-04-21T15:50:01+08:00 wangbomeng server: fix max_seq_len
|
||
06c7852e99e34c71e24e3ef6001cb54c72970e4d 96b4e267e562f44e76ec6c965ee0fa361f0439bf ed62eb33447f993d578e156a3c7e38c91b420ece 2026-04-21T15:06:01+08:00 wangbomeng Merge branch 'dev-ot' into dev
|
||
ed62eb33447f993d578e156a3c7e38c91b420ece 015e988a271573a7c11e2a4b45084e1219d69d50 2026-04-21T15:05:22+08:00 wangbomeng try server-test:not success
|
||
015e988a271573a7c11e2a4b45084e1219d69d50 99962021f2dd1994dbbf90d5712f0ca2633f20f8 2026-04-17T16:56:08+08:00 wangbomeng little change
|
||
96b4e267e562f44e76ec6c965ee0fa361f0439bf 99962021f2dd1994dbbf90d5712f0ca2633f20f8 2026-04-17T15:33:01+08:00 wangbomeng merge dev-ot
|
||
99962021f2dd1994dbbf90d5712f0ca2633f20f8 9af6682ef101c6d006c743c703cf150ac25f466f 2026-04-17T14:48:33+08:00 wangbomeng calrt: add onetoken slice and untile_prefill
|
||
9af6682ef101c6d006c743c703cf150ac25f466f e24f775ee26ea5249d5ff6cd23b8a0a17fe1438d 2026-04-16T10:09:44+08:00 wangbomeng args: add -ca
|
||
e24f775ee26ea5249d5ff6cd23b8a0a17fe1438d a206460e610aef995e36e50b33e8bd41fe477541 2026-04-15T10:01:09+08:00 wangbomeng calrt: add copy_data_to_ibuf for vision model
|
||
a206460e610aef995e36e50b33e8bd41fe477541 d8b2aaa00049978cebac15041f960d75b552f97b 2026-04-14T10:23:43+08:00 wangbomeng calrt: support mixture of prefill_by_ids and prefill_by_embeds
|
||
d8b2aaa00049978cebac15041f960d75b552f97b b6f29ec8140028619efaa50aa6e73146cecf631d 2026-04-13T11:23:44+08:00 wangbomeng server: fix context-shift
|
||
b6f29ec8140028619efaa50aa6e73146cecf631d 6d55eae7e76b768acfcec045b8e84cded8ae3b55 2026-04-13T11:23:31+08:00 wangbomeng server: fix context-shift
|
||
6d55eae7e76b768acfcec045b8e84cded8ae3b55 af8055f5f033a730e7e1fe80185b18b7e9473966 2026-04-10T14:58:56+08:00 wangbomeng add annotations
|
||
af8055f5f033a730e7e1fe80185b18b7e9473966 6f54682df62b8ec6bb4154b99c47b5becda3291f 2026-04-10T14:58:19+08:00 wangbomeng add annotations
|
||
6f54682df62b8ec6bb4154b99c47b5becda3291f 42a181674565411efa5c8b318bc77d6fd084df8a a2d5bf362d6c9a546f9deadf4f8975605a915d33 2026-04-10T09:07:20+08:00 Yunzhe Jia Merge branch 'dev' of http://192.168.10.26:1000/yunzhe/llama.cpp into dev
|
||
42a181674565411efa5c8b318bc77d6fd084df8a bd3a57a11aa008e0b5dc2a30f82fcd3a02587208 2026-04-10T09:06:55+08:00 Yunzhe Jia fix prompt_cache issue
|
||
a2d5bf362d6c9a546f9deadf4f8975605a915d33 bd3a57a11aa008e0b5dc2a30f82fcd3a02587208 2026-04-09T21:01:46+08:00 Bomeng Wang server: comment fixed time
|
||
bd3a57a11aa008e0b5dc2a30f82fcd3a02587208 fcfcdf80b01664a08fb12df879a05df3cc0eba70 74d437ab1456831573def7ca385a74d0ca460c37 2026-04-09T19:42:57+08:00 Yunzhe Jia Merge branch 'dev' of http://192.168.10.26:1000/yunzhe/llama.cpp into dev
|
||
fcfcdf80b01664a08fb12df879a05df3cc0eba70 0cd44929b442860dc0e5f88976108be82350f9dc 2026-04-09T19:41:52+08:00 Yunzhe Jia tmp fix one run >4K problem
|
||
74d437ab1456831573def7ca385a74d0ca460c37 5813c943005d1ede551e67ee98e35aefd210ff22 0cd44929b442860dc0e5f88976108be82350f9dc 2026-04-09T17:44:26+08:00 wangbomeng Merge branch 'dev' of http://192.168.10.26:1000/yunzhe/llama.cpp into dev
|
||
5813c943005d1ede551e67ee98e35aefd210ff22 1f0d5bb0453f93ffa51faf68237c89251c2e0c40 2026-04-09T17:38:57+08:00 wangbomeng add --device-info
|
||
1f0d5bb0453f93ffa51faf68237c89251c2e0c40 b8153b6b8f244b223c9f4c2940ae1f212634519e 2026-04-09T17:38:37+08:00 wangbomeng add --device-info
|
||
0cd44929b442860dc0e5f88976108be82350f9dc b8153b6b8f244b223c9f4c2940ae1f212634519e 2026-04-09T15:08:50+08:00 Bomeng Wang rm calculet0.txt
|
||
b8153b6b8f244b223c9f4c2940ae1f212634519e 82842cfc3fa3d5c004c4540aeff8a81f38509bee 2026-04-09T15:07:02+08:00 wangbomeng server: fix limit tokens to max_seq_len
|
||
82842cfc3fa3d5c004c4540aeff8a81f38509bee 622a22991089c6debd5f4b57738f3df3a0bda8b5 2026-04-09T15:06:26+08:00 wangbomeng server: fix limit tokens to max_seq_len
|
||
622a22991089c6debd5f4b57738f3df3a0bda8b5 5f47a738ba56a37d0ff281a16212f41979568e35 2026-04-09T10:36:31+08:00 wangbomeng server: comment circle print
|
||
5f47a738ba56a37d0ff281a16212f41979568e35 ecf01a17651cd7b1e8b85fd6075e83831463ee23 207ca38c751f7e6c251833eb11e89a2a0bb0fd07 2026-04-09T09:48:20+08:00 Yunzhe Jia Merge branch 'dev' of http://192.168.10.26:1000/yunzhe/llama.cpp into dev
|
||
ecf01a17651cd7b1e8b85fd6075e83831463ee23 99d2078e89305035d87d966cee7987c7d50b3925 2026-04-09T09:47:55+08:00 Yunzhe Jia fix n_parallel problem
|
||
207ca38c751f7e6c251833eb11e89a2a0bb0fd07 99d2078e89305035d87d966cee7987c7d50b3925 2026-04-08T19:14:37+08:00 wangbomeng delet extra printf
|
||
99d2078e89305035d87d966cee7987c7d50b3925 9626239f5a9e568c9975d67dd6745fef90279a87 2026-04-08T19:01:16+08:00 Yunzhe Jia fix buffer resize problem
|
||
9626239f5a9e568c9975d67dd6745fef90279a87 da724f3e2a8af2c2537e3edcff41e9415ac66fef 2026-04-08T17:47:04+08:00 wangbomeng calrt: fix MapBuf
|
||
da724f3e2a8af2c2537e3edcff41e9415ac66fef 18814f9bf1a423ebe8f32d6043c9cbc5b21b0218 2026-04-08T16:58:32+08:00 Bomeng Wang calrt: add MapBuf
|
||
18814f9bf1a423ebe8f32d6043c9cbc5b21b0218 bbee06d6d851e3f84f75b578047967d7e0e4a8dc 2026-04-08T13:55:52+08:00 wangbomeng server : reset inference time
|
||
bbee06d6d851e3f84f75b578047967d7e0e4a8dc e126c21ed7b793b648d63533ea7ee30885f5a82a 2026-04-08T11:29:36+08:00 wangbomeng calrt: add test time print
|
||
e126c21ed7b793b648d63533ea7ee30885f5a82a 5603e050af07173589f3b8850121263d86da01fd 2026-04-01T10:07:12+08:00 wangbomeng sever: fix commit id
|
||
5603e050af07173589f3b8850121263d86da01fd 1e8a16cd0e3ad3c9a0f8746852a0a38c6e52322e 2026-04-01T05:57:55+08:00 wangbomeng calrt: add t_incopy and t_outcopy
|
||
1e8a16cd0e3ad3c9a0f8746852a0a38c6e52322e 893e52bda0fee99bbda1521ea733a760bc4201f1 2026-04-01T10:14:22+08:00 wangbomeng calrt: fix slice and add infer/copy time
|
||
893e52bda0fee99bbda1521ea733a760bc4201f1 398ecf71f5f574037ce33f84dd1576fa164909ad 2026-03-30T11:35:40+08:00 Bomeng Wang delete unnecessary commit id
|
||
398ecf71f5f574037ce33f84dd1576fa164909ad 6d5e83902bdd3142a8ff0d3db19f02a8619e95a0 2026-03-27T02:01:23+08:00 wangbomeng rm llama-server_old.cpp
|
||
6d5e83902bdd3142a8ff0d3db19f02a8619e95a0 e4eaab700333490e80ebefc8d023f6c1adfcc312 2026-03-27T00:57:11+08:00 wangbomeng server: printf calrt commit ID
|
||
e4eaab700333490e80ebefc8d023f6c1adfcc312 aeacc23a883400db0ebffb50e23355ada054da66 2026-03-25T09:03:55+08:00 wangbomeng calrt: prefil to prefill
|
||
aeacc23a883400db0ebffb50e23355ada054da66 cbaa7492663630132adeddccd9fec5e44ffa3336 2026-03-25T08:49:22+08:00 wangbomeng calrt: add slice only on decode
|
||
cbaa7492663630132adeddccd9fec5e44ffa3336 3c08ebf0e442cd730ddffd959ec676f9caf31c82 2026-03-25T03:21:02+08:00 wangbomeng reduce warning
|
||
3c08ebf0e442cd730ddffd959ec676f9caf31c82 613bc2b9a9d7af92efbe42d876ffcdb244f7a00e 2026-03-24T02:43:03+08:00 wangbomeng calrt: comment LOG_ERROR in untile_decode_cpy
|
||
613bc2b9a9d7af92efbe42d876ffcdb244f7a00e 4e194ecc6c4b8ebe4c72d93f6055e3605dbfcefa 2026-03-24T02:30:12+08:00 wangbomeng calrt: fix max_seq_len judgment
|
||
4e194ecc6c4b8ebe4c72d93f6055e3605dbfcefa 16cbf81a731f5839a2f6b0eb9d8539826353748b 2026-03-24T01:59:03+08:00 wangbomeng calrt: fix ubatch.embd cpy
|
||
16cbf81a731f5839a2f6b0eb9d8539826353748b 8b4e17d990c23025148c4abadefe1010a0dd0734 2026-03-23T09:38:21+08:00 wangbomeng calrt: fix model_name of untile_cpy
|
||
8b4e17d990c23025148c4abadefe1010a0dd0734 6dacdd2cf9574214ca272a86674e7bd7ba21c4f0 41e6636ea1fb3884e41fb2023f92b0e7262ef09b 2026-03-23T14:50:27+08:00 wangbomeng merge origin dev
|
||
6dacdd2cf9574214ca272a86674e7bd7ba21c4f0 3e6ad64467771224f072e5bad90e270951ba4cdf 2026-03-23T14:42:28+08:00 wangbomeng calrt: add multimodal
|
||
3e6ad64467771224f072e5bad90e270951ba4cdf b37b7d9756feb3dac2db526eeab38b572a095d47 2026-03-23T14:36:41+08:00 wangbomeng calrt: add multimodal
|
||
41e6636ea1fb3884e41fb2023f92b0e7262ef09b d0a1b2c758440720d42fa108b3444f54593daa73 2026-03-12T16:36:07+08:00 Bomeng Wang server: correct the max_seq_len judgment
|
||
d0a1b2c758440720d42fa108b3444f54593daa73 b37b7d9756feb3dac2db526eeab38b572a095d47 2026-03-12T16:10:28+08:00 Bomeng Wang calrt: comment out dump
|
||
b37b7d9756feb3dac2db526eeab38b572a095d47 07817cefc14fa359dcecad0630defd3610f8f5c8 7ddafbdcb9426ae3af016925b8b596f11e8742d0 2026-03-10T16:32:31+08:00 Yunzhe Jia Merge branch 'dev' of http://192.168.10.26:1000/yunzhe/llama.cpp into dev
|
||
07817cefc14fa359dcecad0630defd3610f8f5c8 b2c45d8ddefb7ec09435231c8ec5b736fabf9c91 2026-03-10T16:32:07+08:00 Yunzhe Jia fix kv issue
|
||
7ddafbdcb9426ae3af016925b8b596f11e8742d0 059009eeaf20a569abfbac5eb37dad3bb9376428 2026-03-10T07:13:04+08:00 wangbomeng calrt: commented out cal_ctx->encode(batch)
|
||
059009eeaf20a569abfbac5eb37dad3bb9376428 1b073661d1311b5300173993b9f31ee7241d8606 2026-03-10T07:04:22+08:00 wangbomeng calrt: add multimodel
|
||
1b073661d1311b5300173993b9f31ee7241d8606 6295273c5866b9249bacdedbfc3a31877e6e6a85 b2c45d8ddefb7ec09435231c8ec5b736fabf9c91 2026-03-10T06:58:35+08:00 wangbomeng Merge branch 'dev' of http://192.168.20.229:1000/yunzhe/llama.cpp into dev
|
||
6295273c5866b9249bacdedbfc3a31877e6e6a85 c6ea3f0d2468e3b8eab5de80bacee7fe4c9dd3d7 2026-03-10T06:54:41+08:00 wangbomeng arg: update hf_repo url to https://download.calculet.tech:9443
|
||
b2c45d8ddefb7ec09435231c8ec5b736fabf9c91 c6ea3f0d2468e3b8eab5de80bacee7fe4c9dd3d7 2026-03-09T15:03:26+08:00 Bomeng Wang clart: past_kv_len = n_tokens
|
||
c6ea3f0d2468e3b8eab5de80bacee7fe4c9dd3d7 f8fba66b657c51a1df1efc7267bfea9f6140cebf 2026-03-09T03:26:47+08:00 wangbomeng CMakeLists: STATIC to SHARED
|
||
f8fba66b657c51a1df1efc7267bfea9f6140cebf db4a61d2234c2f457b42ecc3f7219a1e1a35508a 2026-03-06T15:06:36+08:00 Yunzhe Jia fix calrt_install include path
|
||
db4a61d2234c2f457b42ecc3f7219a1e1a35508a e672e02ca1a582f9e3a9d011d9e1b4fea4942a25 2026-03-06T06:59:58+08:00 Yunzhe Jia adapt to host-rt install structure
|
||
e672e02ca1a582f9e3a9d011d9e1b4fea4942a25 27b82f64252779ad160a3fa76bd378c39421de36 7a3a9a0e76bb5f530beafcfe52e87e388e93117a 2026-03-06T11:47:09+08:00 Yunzhe Jia Merge branch 'dev' of http://192.168.10.26:1000/yunzhe/llama.cpp into dev
|
||
27b82f64252779ad160a3fa76bd378c39421de36 0bd95719fed2f0247c8d0199dd897d24157dfae6 2026-03-06T11:42:45+08:00 Yunzhe Jia adapt to calrt_objs target
|
||
7a3a9a0e76bb5f530beafcfe52e87e388e93117a 583c387efaf7bc030574e554088de79d09a27fbd 2026-03-05T07:24:59+08:00 wangbomeng calrt_infer: fix past_kv_len
|
||
583c387efaf7bc030574e554088de79d09a27fbd 0edc14a60a2d5596f71d728c655e3523da924d89 2026-02-27T02:24:56+08:00 wangbomeng caserver: lim generated tokens to n_ctx_per_seq
|
||
0edc14a60a2d5596f71d728c655e3523da924d89 aa0a17d719326a63e0d6fea60aa0ced44e7abc33 2026-02-27T01:57:39+08:00 wangbomeng caserver: limit generated tokens to n_ctx_per_seq
|
||
aa0a17d719326a63e0d6fea60aa0ced44e7abc33 fa332e773da681f3e77d6ffe83a87edb557f758b 2026-02-27T01:40:11+08:00 wangbomeng calrt: add bf16_to_fp32
|
||
fa332e773da681f3e77d6ffe83a87edb557f758b aa54a7c637f2c92d924a5f35386975b1c27de898 2026-02-27T01:31:42+08:00 wangbomeng Add bf16_to_fp32
|
||
aa54a7c637f2c92d924a5f35386975b1c27de898 84f2c0b5f334a7ab77347c6779b6027a951d0ff0 2026-02-27T01:30:36+08:00 wangbomeng Add bf16_to_fp32
|
||
84f2c0b5f334a7ab77347c6779b6027a951d0ff0 0bd95719fed2f0247c8d0199dd897d24157dfae6 2026-02-09T01:46:36+08:00 wangbomeng add bf16_to_fp32 in untile_decode/prefill_cpy
|
||
0bd95719fed2f0247c8d0199dd897d24157dfae6 365eb0b719e30b0f9e348d854f11c9c3f954a96e 2026-02-07T10:25:06+08:00 Yunzhe Jia fix get_model_by_name
|
||
365eb0b719e30b0f9e348d854f11c9c3f954a96e 886cd1fb3b217efcf51c46209fcb694022346797 2026-02-07T09:05:17+08:00 Yunzhe Jia comment out pos buffer file
|
||
886cd1fb3b217efcf51c46209fcb694022346797 9328f21378275f2354a19c3af907332b216ab492 2026-02-06T17:36:54+08:00 Yunzhe Jia fix compile problem
|
||
9328f21378275f2354a19c3af907332b216ab492 49fea70a20f550bfdb59d0ffe041b54919c22447 2026-01-28T15:32:36+08:00 Yunzhe Jia add prefill-only mode
|
||
49fea70a20f550bfdb59d0ffe041b54919c22447 74a118df31b6a4acc8876b2a2db90d2395cc6ec3 2026-01-28T09:11:16+08:00 Yunzhe Jia adapt to new calbin test_case
|
||
74a118df31b6a4acc8876b2a2db90d2395cc6ec3 98e8f9acfe87cd93646482cb2d942b8a132f0852 2026-01-13T14:32:47+08:00 Yunzhe Jia sync with calrt update
|
||
98e8f9acfe87cd93646482cb2d942b8a132f0852 8c8152a04036bb0d4756a4e08e54f514dae4e64d 2025-12-10T00:20:57+08:00 Yunzhe Jia add timestamp
|
||
8c8152a04036bb0d4756a4e08e54f514dae4e64d 5d2387931528fbe04aa8d66de935147efb65ca4d 2025-12-09T14:40:04+08:00 Yunzhe Jia adapt to calrt csr
|
||
5d2387931528fbe04aa8d66de935147efb65ca4d e6285a748657add8d974b78ca920c5b41753ee12 2025-12-08T16:40:34+08:00 Yunzhe Jia add timestamp for one decode process
|
||
e6285a748657add8d974b78ca920c5b41753ee12 6b1ecdbbe3cc72ed8888be38f9e934cfd79b9fbb 2025-11-27T16:58:35+08:00 Yunzhe Jia adapt to calrt
|
||
6b1ecdbbe3cc72ed8888be38f9e934cfd79b9fbb 240d9bb75241c91896afe125f2fc74698a7395f3 2025-11-25T17:06:14+08:00 Yunzhe Jia add kv_tensor for pld test
|
||
240d9bb75241c91896afe125f2fc74698a7395f3 7eeb241aad429a68fb8cfa3ba85d000fb2348cd2 2025-11-24T12:03:00+08:00 Yunzhe Jia adapt to calrt modification
|
||
7eeb241aad429a68fb8cfa3ba85d000fb2348cd2 dfb6250031a25058dbc3757e8a7ec75d90dcb48a 2025-11-21T09:20:01+08:00 Yunzhe Jia fix byte_size error
|
||
dfb6250031a25058dbc3757e8a7ec75d90dcb48a e31bce25e85dc7fdf0a0e69fd0e58ba4a2ffaf3e 2025-11-18T10:28:00+08:00 Yunzhe Jia add elf in CMakelist
|
||
e31bce25e85dc7fdf0a0e69fd0e58ba4a2ffaf3e 6444e227c36df16199c40d8cead2244fe014f377 2025-11-07T11:52:17+08:00 Yunzhe Jia dump src tensor
|
||
6444e227c36df16199c40d8cead2244fe014f377 d81d7b190ff5f21325167936496dfe5ab48b922e 2025-11-07T09:19:11+08:00 Yunzhe Jia test golden case
|
||
d81d7b190ff5f21325167936496dfe5ab48b922e e492f5dc8cec17529f53ef2bbdf7b502107f8148 1f5accb8d0056e6099cd5b772b1cb787dd590a13 2025-11-04T14:43:32+08:00 Yunzhe Jia Merge branch 'master' into dev
|
||
e492f5dc8cec17529f53ef2bbdf7b502107f8148 2e05afd22b3d77c7013b11eee88e88b17188f21a 2025-11-04T14:09:59+08:00 Yunzhe Jia redo untile logic
|
||
1f5accb8d0056e6099cd5b772b1cb787dd590a13 2759ccdb4adc8568add4316780d5e675519b0775 2025-11-04T05:04:59Z Noah Fix garbled output with REPACK at high thread counts (#16956)
|
||
2759ccdb4adc8568add4316780d5e675519b0775 c5023daf607c578d6344c628eb7da18ac3d92d32 2025-11-04T10:53:48+08:00 Aman Gupta CUDA: avoid mul + bias fusion when doing fusion (#16935)
|
||
c5023daf607c578d6344c628eb7da18ac3d92d32 e7da30b584dc1f2ee0414c4a1298ce64eef97e8d 2025-11-03T11:47:57-08:00 lhez opencl: support imrope (#16914)
|
||
e7da30b584dc1f2ee0414c4a1298ce64eef97e8d ed8aa63320393512bdcfe4b05b5ae01ba91888e1 2025-11-03T18:53:26+01:00 Aleksander Grygier fix: Viewing multiple PDF attachments (#16974)
|
||
ed8aa63320393512bdcfe4b05b5ae01ba91888e1 48bd26501b08a3f0bff1249db47f313641f7bebb 2025-11-03T18:01:59+01:00 Daniel Bevenius model-conversion : pass config to from_pretrained (#16963)
|
||
48bd26501b08a3f0bff1249db47f313641f7bebb 622cd010ff4a65bef67edbf2f9bf4707c01f98f7 2025-11-03T15:38:23+02:00 Georgi Gerganov server : add props.model_alias (#16943)
|
||
622cd010ff4a65bef67edbf2f9bf4707c01f98f7 070ff4d5356083d60b807bb34d36b31c3653a29e 2025-11-03T14:29:11+01:00 theo77186 ggml: CUDA: add head size 72 for flash-attn (#16962)
|
||
070ff4d5356083d60b807bb34d36b31c3653a29e bf7b0c9725ed0c406c5debaea022ec7258430634 2025-11-03T11:11:18+01:00 Xuan-Son Nguyen mtmd: add --image-min/max-tokens (#16921)
|
||
bf7b0c9725ed0c406c5debaea022ec7258430634 fcfce040e816c542f374ad51461b3561d73c4bc9 2025-11-03T10:25:55+01:00 Xuan-Son Nguyen mtmd: pad mask for qwen2.5vl (#16954)
|
||
fcfce040e816c542f374ad51461b3561d73c4bc9 ee3a5a10adf9e83722d1914dddc56a0623ececaf 2025-11-03T14:40:02+08:00 Jinyang He ggml : LoongArch fixes (#16958)
|
||
ee3a5a10adf9e83722d1914dddc56a0623ececaf 7e994168b1ccc12337ba8de939c4fd466107c1fb 2025-11-03T05:33:56Z Olivier Chafik sync: minja (glm 4.6 & minmax m2 templates) (#16949)
|
||
7e994168b1ccc12337ba8de939c4fd466107c1fb bcfa87622ae46be6345a8e3dfdbdc5ba5414042b 2025-11-03T03:35:33+02:00 shani-f SYCL: optimized repeat_back kernel (3× fewer asm instructions, 2× faster)Feature/sycl repeat back opt (#16869)
|
||
bcfa87622ae46be6345a8e3dfdbdc5ba5414042b a2054e3a8ff0da3978a4acc18c349ff58554d336 2025-11-03T00:41:08+01:00 Sascha Rogmann feat(webui): improve LaTeX rendering with currency detection (#16508)
|
||
a2054e3a8ff0da3978a4acc18c349ff58554d336 dd5286805004db1f9ac3176a1cbbfe373bdda0f8 2025-11-03T04:40:30+05:30 Shagun Bera test-backend-ops : fix segfault in moe-expert-reduce test in support mode and coverage (#16936)
|
||
dd5286805004db1f9ac3176a1cbbfe373bdda0f8 6b9a52422bac0f50dd8f1f8386744fa3ce9783bf 2025-11-02T23:11:21+01:00 Sigbjørn Skjæret ci : disable failing riscv cross build (#16952)
|
||
6b9a52422bac0f50dd8f1f8386744fa3ce9783bf 2f966b8ed87514e74bb96592217226cb6a6974dd 2025-11-02T13:08:04-08:00 Zhiyong Wang model: add Janus Pro for image understanding (#16906)
|
||
2f966b8ed87514e74bb96592217226cb6a6974dd cd5e3b57541ecc52421130742f4d89acbcf77cd4 2025-11-02T22:21:48+02:00 Georgi Gerganov clip : use FA (#16837)
|
||
cd5e3b57541ecc52421130742f4d89acbcf77cd4 87c9efc3b297b8a498716b1db3d061842e6fc85b 2025-11-02T18:14:04+02:00 Georgi Gerganov server : support unified cache across slots (#16736)
|
||
87c9efc3b297b8a498716b1db3d061842e6fc85b 76af40aaaad78c42faecd8016a88362c788b84b0 2025-11-02T08:56:28-06:00 Aldehir Rojas common : move gpt-oss reasoning processing to init params (#16937)
|
||
76af40aaaad78c42faecd8016a88362c788b84b0 7db35a7958a943be1693879f42d166f152613979 2025-11-02T10:28:37+01:00 Adrian Lundberg docs: remove llama_sampler_accept reference in sampling sample usage (#16920)
|
||
7db35a7958a943be1693879f42d166f152613979 a864132ba546b6385922226480ee4e392aaa065c 2025-11-02T08:42:57+05:30 mnehete32 CUDA: add FLOOR, CEIL, ROUND, TRUNC unary ops (#16917)
|
||
a864132ba546b6385922226480ee4e392aaa065c d38d9f0877a5872daa3c5f06fb9a86376bf15d50 2025-11-02T08:48:46+08:00 Aaron Teo devops: fix failing s390x docker build (#16918)
|
||
d38d9f0877a5872daa3c5f06fb9a86376bf15d50 7fd205a8e8832ead273f62c5fd81d2b8aa910535 2025-11-02T08:48:23+08:00 Aaron Teo ggml: add s390x cpu-feats (#16774)
|
||
7fd205a8e8832ead273f62c5fd81d2b8aa910535 2f68ce7cfd20e9e7098514bf730e5389b7bba908 2025-11-02T00:15:31+02:00 Georgi Gerganov scripts : add script to bench models (#16894)
|
||
2f68ce7cfd20e9e7098514bf730e5389b7bba908 e4a71599e5846110159955dec0008eb4aa24222b 2025-11-01T19:49:51+01:00 Pascal webui: auto-refresh /props on inference start to resync model metadata (#16784)
|
||
e4a71599e5846110159955dec0008eb4aa24222b dd5e8cab512c7752392b6e51a6f118f348fb3f16 2025-11-01T17:14:54+01:00 Pascal webui: add HTML/JS preview support to MarkdownContent with sandboxed iframe (#16757)
|
||
dd5e8cab512c7752392b6e51a6f118f348fb3f16 cf659bbb8ef9eb048e5153a27cf787fd83c05560 2025-11-01T16:52:17+01:00 Adrien Gallouët vendor : update cpp-httplib to 0.27.0 (#16846)
|
||
cf659bbb8ef9eb048e5153a27cf787fd83c05560 d8b860a219c2415faac8cc0e50b48b4aa11e3b64 2025-11-01T15:51:36+01:00 Xuan-Son Nguyen mtmd: refactor preprocessing + support max/min pixels (#16878)
|
||
d8b860a219c2415faac8cc0e50b48b4aa11e3b64 1ae74882f8f6755e44dff8f23f3abdc5b53ab7c1 2025-11-01T15:35:57+01:00 Aleksander Grygier Add a setting to display message generation statistics (#16901)
|
||
1ae74882f8f6755e44dff8f23f3abdc5b53ab7c1 961660b8c395afb8903f61ace69396a158ea0cb4 2025-11-01T15:02:57+01:00 Jaromír Hradílek webui: recognize AsciiDoc files as valid text files (#16850)
|
||
961660b8c395afb8903f61ace69396a158ea0cb4 74fef4129fa58f6277c243777a4894371479dbad 2025-11-01T11:01:42+01:00 Sigbjørn Skjæret common : allow --system-prompt-file for diffusion-cli (#16903)
|
||
74fef4129fa58f6277c243777a4894371479dbad 5d8bb900bc7daa84bfa7bb1d25ab7e32394919f3 2025-11-01T08:55:25+01:00 Sigbjørn Skjæret codeowners : update after refactor (#16905)
|
||
5d8bb900bc7daa84bfa7bb1d25ab7e32394919f3 2e76e013600cb0d51ccf158571ca1d0502952a07 2025-11-01T00:52:14-05:00 Jeff Bolz vulkan: Fix multi_add invalid descriptor usage (#16899)
|
||
2e76e013600cb0d51ccf158571ca1d0502952a07 d3dc9dd898be805c23a408cc36daed5b3bf29221 2025-11-01T00:45:28-05:00 Jeff Bolz vulkan: fuse mul_mat+add and mul_mat_id+add_id (#16868)
|
||
d3dc9dd898be805c23a408cc36daed5b3bf29221 bea04522ff1a0d8559ccfd353aa018dcfbb608cc 2025-11-01T06:13:26+01:00 Oliver Simons CUDA: Remove unneded bias/gate dims in fused mmvq (#16858)
|
||
bea04522ff1a0d8559ccfd353aa018dcfbb608cc 0de0a01576772032008a689afc4d7c80685074c4 2025-10-31T23:40:23+01:00 Piotr Wilkin (ilintar) refactor : llama-model.cpp (#16252)
|
||
0de0a01576772032008a689afc4d7c80685074c4 e58d585604bbbbbc24f8149effe444621b5191c6 2025-10-31T21:20:47+01:00 Piotr Wilkin (ilintar) model : Minimax M2 (#16831)
|
||
e58d585604bbbbbc24f8149effe444621b5191c6 31c511a968348281e11d590446bb815048a1e912 2025-10-31T21:20:07+01:00 Giuseppe Scrivano model : add Granite Hybrid nano types (#16896)
|
||
31c511a968348281e11d590446bb815048a1e912 6d39015a744b47bb7643ea73f412e6d64677f305 2025-10-31T15:57:19+01:00 Johannes Gäßler CUDA: Volta tensor core support for MMF (#16843)
|
||
6d39015a744b47bb7643ea73f412e6d64677f305 4146d6a1a6228711a487a1e3e9ddd120f8d027d7 2025-10-31T16:25:50+02:00 Georgi Gerganov sync : ggml
|
||
4146d6a1a6228711a487a1e3e9ddd120f8d027d7 8da3c0e200a586f768ada6f38745acb01380174c 2025-10-31T20:05:07+08:00 Aman Gupta CUDA: add expert reduce kernel (#16857)
|
||
8da3c0e200a586f768ada6f38745acb01380174c c22473b580807929fd9e3a3344a48e8cfbe6c88f 2025-10-31T13:50:33+02:00 Georgi Gerganov batch : fix consistency checks for the input positions (#16890)
|
||
c22473b580807929fd9e3a3344a48e8cfbe6c88f 0f715b4e759acceccb9f437cfd2a988fff85514a 2025-10-31T10:54:19+02:00 Georgi Gerganov server : don't print user inputs to console (#16871)
|
||
0f715b4e759acceccb9f437cfd2a988fff85514a d2d931f173b8a736b08999436e9259aafddec718 2025-10-31T09:51:26+01:00 Daniel Bevenius server : fix typos in server.cpp comments [no ci] (#16883)
|
||
d2d931f173b8a736b08999436e9259aafddec718 2976b0374d36609b0429dd6ce48807e2ad39a7c2 2025-10-31T02:34:47-05:00 Jeff Bolz vulkan: disable spirv-opt for rope shaders (#16872)
|
||
2976b0374d36609b0429dd6ce48807e2ad39a7c2 d2a2673dd1c86a01ab010e18b13b8bb959968c48 2025-10-31T16:18:59+09:00 Masato Nakasaka vulkan: Fix crash when FP16 mul_mat accumulation is not supported (#16796)
|
||
d2a2673dd1c86a01ab010e18b13b8bb959968c48 13002a08960e51a76c4d696165b5d7638d2f2b99 2025-10-31T08:14:49+01:00 Ruben Ortlam vulkan: fix shmem overrun in mmq id shader (#16873)
|
||
2e05afd22b3d77c7013b11eee88e88b17188f21a c20c0a5c0c44179f5267a262546624854b1203d1 2025-10-31T14:06:06+08:00 Yunzhe Jia add pcie support
|
||
13002a08960e51a76c4d696165b5d7638d2f2b99 6eb208d17ea29bb60295d9a2b5e7122dfb8f4b55 2025-10-31T12:46:31+08:00 l3utterfly ggml-hexagon: respect input size when getting/setting tensor data (#16836)
|
||
6eb208d17ea29bb60295d9a2b5e7122dfb8f4b55 9984cbb61d12491d604484fb38b678fa15064061 2025-10-31T00:34:27+01:00 Sigbjørn Skjæret ci : enable free-disk-space on cuda docker build (#16877)
|
||
9984cbb61d12491d604484fb38b678fa15064061 ce18efeaf1234bd08f25bc08f88ef95cf5d9e51f 2025-10-30T16:00:20-07:00 lhez opencl: fix boundary handling for mul_mm (#16875)
|
||
ce18efeaf1234bd08f25bc08f88ef95cf5d9e51f 16724b5b6836a2d4b8936a5824d2ff27c52b4517 2025-10-30T23:15:03+01:00 RodriMora convert : update transformers requirements (#16866)
|
||
16724b5b6836a2d4b8936a5824d2ff27c52b4517 b52edd25586fabb70f0c21b274473b307cf14499 2025-10-30T14:22:23-04:00 chansikpark server : bump request URI max length to 32768 (#16862)
|
||
b52edd25586fabb70f0c21b274473b307cf14499 517b7170e1a4d733583c4b07c5b7a49acc05911c 2025-10-30T18:42:57+02:00 Georgi Gerganov server : remove n_past (#16818)
|
||
517b7170e1a4d733583c4b07c5b7a49acc05911c 835e918d8428f5119927d7150bf5a26176dedda0 2025-10-30T09:06:13-07:00 Max Krasnyansky cpu: introduce chunking for repack matmuls and enable matmul-id chunking on ARM64 (#16833)
|
||
835e918d8428f5119927d7150bf5a26176dedda0 d261223d24e97f2df50220e4a5b7f0adb69bba81 2025-10-30T21:17:31+05:30 Shagun Bera common: fix typo in cli help text (#16864)
|
||
d261223d24e97f2df50220e4a5b7f0adb69bba81 dcca0d3ab840ebe9b2ccd4719033d408eeb758d7 2025-10-30T23:19:14+08:00 JJJYmmm model: add support for qwen3vl series (#16780)
|
||
dcca0d3ab840ebe9b2ccd4719033d408eeb758d7 bacddc049a00786df44e682262f6e298742bfbc3 2025-10-30T05:26:05-07:00 Max Krasnyansky cpu: introduce chunking for flash attention (#16829)
|
||
bacddc049a00786df44e682262f6e298742bfbc3 229bf686287d18f82c44e89888cc662145ecfdb4 2025-10-30T07:18:50-04:00 Tianyue-Zhao model: Add support for CogVLM model (#15002)
|
||
229bf686287d18f82c44e89888cc662145ecfdb4 d7395115baf395b75a73a17b0b796e746e468da9 2025-10-30T08:56:28+01:00 Sigbjørn Skjæret cuda : fix argsort with 64k+ rows (#16849)
|
||
d7395115baf395b75a73a17b0b796e746e468da9 052df28b0e3ca0398e5928c61c7de40254317894 2025-10-30T14:30:58+08:00 Jan Boon llama : use std::abs instead of abs (#16853)
|
||
052df28b0e3ca0398e5928c61c7de40254317894 8b11deea4663f29d3e042ce1056ba643264cd5f1 2025-10-30T01:27:41-05:00 Jeff Bolz vulkan: Handle argsort with a large number of rows (#16851)
|
||
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)
|
||
3464bdac37027c5e9661621fc75ffcef3c19c6ef e3af5563bd049141e036b50f843196db33d23e97 2025-10-29T20:11:39+01:00 Xuan-Son Nguyen llama: fix ASAN error with M-RoPE (#16848)
|
||
e3af5563bd049141e036b50f843196db33d23e97 10fcc41290e233788f5a4215314156e8e023eb92 2025-10-29T18:09:18+01:00 Xuan-Son Nguyen llama: store mrope data in KV cell (#16825)
|
||
10fcc41290e233788f5a4215314156e8e023eb92 bcf5bda6f5df559565d11d7c8e8295c1159a85ec 2025-10-29T08:44:29-05:00 Jeff Bolz vulkan: Update topk_moe fusion to handle gpt's late softmax (#16656)
|
||
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)
|
||
e41bcce8f0b53032a1fed275cd253e931c041cf6 144a4ce824b6bd0e48d62009d10cae1daf5308db 2025-10-29T21:11:53+08:00 Aman Gupta CUDA: use fastdiv in set-rows (#16834)
|
||
144a4ce824b6bd0e48d62009d10cae1daf5308db f549b0007dbdd683215820f7229ce180a12b191d 2025-10-29T14:09:50+01:00 Sigbjørn Skjæret vendor : sync minja (#16500)
|
||
f549b0007dbdd683215820f7229ce180a12b191d 9a3ea685b937c0f0cbfda2e50004ea54bf187512 2025-10-29T03:53:04-05:00 Jeff Bolz vulkan: Call ggml_vk_buffer_write_2d from ggml_vk_buffer_copy (#16793)
|
||
9a3ea685b937c0f0cbfda2e50004ea54bf187512 338074c383c81366320d176d83b94b0a567ee0c2 2025-10-29T15:55:06+08:00 Aman Gupta CUDA: Fix bug in topk-moe for gpt-oss (#16821)
|
||
338074c383c81366320d176d83b94b0a567ee0c2 851553ea6b24cb39fd5fd188b437d777cb411de8 2025-10-29T08:14:39+02:00 YaelLogic sycl: add RMS_NORM_BACK operation support (#16808)
|
||
851553ea6b24cb39fd5fd188b437d777cb411de8 85a7d8677bf2200981e52f744a21d5267964ffcf 2025-10-28T21:10:28+02:00 YaelGitAccount cuda: add SET operation support (#16804)
|
||
85a7d8677bf2200981e52f744a21d5267964ffcf a8ca18b4b815a2abdbecb958ee5f4c542d69aac7 2025-10-28T20:19:44+02:00 Georgi Gerganov memory : remove KV cache size padding (#16812)
|
||
a8ca18b4b815a2abdbecb958ee5f4c542d69aac7 8284efc35c909217ee1b9a06903245d808ac2283 2025-10-28T19:41:43+02:00 Georgi Gerganov llama-bench : clarify benchmarked parts of the computation (#16823)
|
||
8284efc35c909217ee1b9a06903245d808ac2283 1c1409e13169d0ced4b1b4d39fb5b268b7525091 2025-10-28T23:16:20+08:00 l3utterfly initialise buffer.device in ggml_hexagon_session (#16816)
|
||
1c1409e13169d0ced4b1b4d39fb5b268b7525091 7a0e900e3615fa46c074a7fdf900b47d3c0a1c7e 2025-10-28T03:51:41-07:00 Sam Malayek embedding: add raw option for --embd-output-format (#16541)
|
||
7a0e900e3615fa46c074a7fdf900b47d3c0a1c7e 280d97be9660e7a5feaa28a6e7a299bc73dd83fc 2025-10-28T11:23:54+01:00 Johannes Gäßler llama: consistent ctx <-> buf order for KV cache (#16746)
|
||
280d97be9660e7a5feaa28a6e7a299bc73dd83fc 3479efd112b3910d2d008ad08fe4cb4895605903 2025-10-28T03:37:52-05:00 Aldehir Rojas grammar : support array references in json schema (#16792)
|
||
3479efd112b3910d2d008ad08fe4cb4895605903 463bbf20bfbe0da10ac58984d4d62bed5a49362a 2025-10-28T10:54:53+08:00 Chenguang Li CANN: Improve device ID handling and aclnnArange checks (#16752)
|
||
463bbf20bfbe0da10ac58984d4d62bed5a49362a ad8d36beffd791db10c94eb9e964afb891e3ca55 2025-10-28T10:31:21+08:00 Aman Gupta CUDA: add unused vars to mmvf and mmvq (#16807)
|
||
ad8d36beffd791db10c94eb9e964afb891e3ca55 c053e18a66dd95dc340aa61317877c2a41d4e3cf 2025-10-28T03:50:33+02:00 tamarPal sycl: add SSM_CONV operation support (#16800)
|
||
c053e18a66dd95dc340aa61317877c2a41d4e3cf e1ab0848037c9f9bfe68b3e1cee9ee375e1018a3 2025-10-27T18:54:01-04:00 Yuri Khrustalev chat: Add LFM2 tool handling (#16763)
|
||
e1ab0848037c9f9bfe68b3e1cee9ee375e1018a3 5a4ff43e7dd049e35942bc3d12361dab2f155544 2025-10-27T23:12:16+01:00 Xuan-Son Nguyen mtmd : fix idefics3 preprocessing (#16806)
|
||
5a4ff43e7dd049e35942bc3d12361dab2f155544 10640e31aab0819f31c1e1f2d008b019ee737232 2025-10-27T13:51:28-07:00 Diego Devesa llama : disable pipeline parallelism if compute buffer allocation fails (#16748)
|
||
10640e31aab0819f31c1e1f2d008b019ee737232 80d28f104c0c3e61c11d6af073642b246a9fc19c 2025-10-27T21:50:22+01:00 Acly ggml : fix interpolate with align-corners and ne=1 (#16700)
|
||
80d28f104c0c3e61c11d6af073642b246a9fc19c c55d53acec864f64afa1ba92972203dce1bf88f5 2025-10-27T21:39:49+01:00 Johannes Gäßler HIP: fix AMDGPU_TARGETS, update documentation (#16803)
|
||
c55d53acec864f64afa1ba92972203dce1bf88f5 945501f5ea4b8ca56b181ecb035e9ee3fb31f432 2025-10-27T16:02:58+01:00 Xuan-Son Nguyen model : add LightOnOCR-1B model (#16764)
|
||
945501f5ea4b8ca56b181ecb035e9ee3fb31f432 75cbdd3fce38ea12d50cd19e73a069aa5dbbd5fa 2025-10-27T09:17:31+01:00 Johannes Gäßler llama: fix leaked buffers for mmap + split files (#16765)
|
||
75cbdd3fce38ea12d50cd19e73a069aa5dbbd5fa 2b9bd9bf4e759c05db629ec1c391dc8aeaa71887 2025-10-27T09:25:10+08:00 Aman Gupta test-backend-ops: print failed tests at the end (#16785)
|
||
2b9bd9bf4e759c05db629ec1c391dc8aeaa71887 59fc1ec8e83b14354c1a3a8acf8c5c2cbf9af42f 2025-10-27T03:20:24+02:00 tamarPal sycl: add ROLL operation support (#16665)
|
||
59fc1ec8e83b14354c1a3a8acf8c5c2cbf9af42f 75d33b9302f84a5b89f82205d2bcd8def5a64e0a 2025-10-27T03:19:50+02:00 shani-f sycl: add REPEAT_BACK operation support (#16734)
|
||
75d33b9302f84a5b89f82205d2bcd8def5a64e0a 3470a5c891dcc94363e492a3760af92b6b07241c 2025-10-27T09:06:16+08:00 Aman Gupta CUDA: support for weight clamp in top-k norm (#16702)
|
||
3470a5c891dcc94363e492a3760af92b6b07241c bd562fe4f7bd55625511d5f9d639c4fb1db1d440 2025-10-26T23:19:03+01:00 Acly ggml-alloc : make gallocr prefer chunks that allow memory reuse (#16788)
|
||
bd562fe4f7bd55625511d5f9d639c4fb1db1d440 bbac6a26b2bd7f7c1f0831cb1e7b52734c66673b 2025-10-26T21:31:41+01:00 Sigbjørn Skjæret cuda : use fast copy when src and dst are of different type and contiguous (#16789)
|
||
bbac6a26b2bd7f7c1f0831cb1e7b52734c66673b 73a48c9790d320476b3e5ef75bda09f2f8269e6e 2025-10-27T02:13:31+08:00 leejet ggml: fix cuda kernel launch configuration for k_compute_batched_ptrs to support large batch (#16744)
|
||
73a48c9790d320476b3e5ef75bda09f2f8269e6e f696428ce8e4d16c17acbffeaa7feac3b0fb9061 2025-10-26T17:21:23+01:00 Sigbjørn Skjæret convert : enable expert group selection for all models with it (#16691)
|
||
f696428ce8e4d16c17acbffeaa7feac3b0fb9061 7cce4f8158f0c4c88d8dadd4c23d33938127b897 2025-10-26T17:20:32+01:00 Sigbjørn Skjæret graph : add clamping to ffn_moe_weights_sum to avoid div-by-zero (#16655)
|
||
7cce4f8158f0c4c88d8dadd4c23d33938127b897 8d8862829cd770fc9c0c7a726ed162de824ff5ea 2025-10-26T16:08:52+01:00 Sigbjørn Skjæret model : set res->t_embd in SmallThinker models (#16782)
|
||
8d8862829cd770fc9c0c7a726ed162de824ff5ea f77c13b91f4d25754b6a0b857f98a6bc922a0aa7 2025-10-26T14:01:20+02:00 amirai21 docs : add Jamba to Text-only models list (#16778)
|
||
f77c13b91f4d25754b6a0b857f98a6bc922a0aa7 3cfa9c3f125763305b4226bc032f1954f08990dc 2025-10-26T19:28:04+08:00 Aman Gupta CUDA: General GEMV fusion (#16715)
|
||
3cfa9c3f125763305b4226bc032f1954f08990dc 5d195f17bc60eacc15cfb929f9403cf29ccdf419 2025-10-26T06:37:38+02:00 Gilad S. vulkan: deduplicate Microsoft Direct3D12 devices (#16689)
|
||
5d195f17bc60eacc15cfb929f9403cf29ccdf419 226f295f4dd92ad714533adc5497afed5fa88bb8 2025-10-25T20:41:36+02:00 Galunid convert : handle mmproj filename/path properly (#16760)
|
||
226f295f4dd92ad714533adc5497afed5fa88bb8 f90b4a8efe4466215c51155a5c22c1ae6207da23 2025-10-25T19:26:27+09:00 Shunta Saito model : set res->t_embd in PLaMo2 models (#16766)
|
||
f90b4a8efe4466215c51155a5c22c1ae6207da23 8423d019318b446640bb620e4ce80066d8530f05 2025-10-25T10:59:54+02:00 Giuseppe Scrivano vulkan: delete dead code (#16732)
|
||
8423d019318b446640bb620e4ce80066d8530f05 5cca2542ac3f3f86831d32bce744d08fc2b353b0 2025-10-25T00:04:12-05:00 Jeff Bolz vulkan: Optimize SSM_SCAN (#16645)
|
||
5cca2542ac3f3f86831d32bce744d08fc2b353b0 55945d2ef51b93821d4b6f4a9b994393344a90db 2025-10-24T20:52:00-04:00 compilade convert : avoid dequantizing mxfp4 for GPT-OSS (#16756)
|
||
55945d2ef51b93821d4b6f4a9b994393344a90db 0bcb40b48c6fc6f17ba9672625e526ab2574344b 2025-10-25T03:39:37+08:00 leejet ggml: fix CUDA grid launch condition for large block_nums.y in binbcast (#16742)
|
||
0bcb40b48c6fc6f17ba9672625e526ab2574344b 69e9ff010309a1155d704cf9320bdb3aaf4160ca 2025-10-24T20:46:19+08:00 Aman Gupta CUDA: use CUB for arbitary size argsort (#16754)
|
||
69e9ff010309a1155d704cf9320bdb3aaf4160ca 5a91109a5d7dab5d7adc40bedb397ede99a705b1 2025-10-24T14:10:29+02:00 Florian Badie webui: support q URL parameter (#16728)
|
||
5a91109a5d7dab5d7adc40bedb397ede99a705b1 f8f071faddf32ea09f4234edb6e809b380a9ee26 2025-10-24T12:02:02+02:00 Daniel Bevenius model-conversion : add trust_remote_code for orig model run [no ci] (#16751)
|
||
f8f071faddf32ea09f4234edb6e809b380a9ee26 0bf47a1dbba4d36f2aff4e8c34b06210ba34e688 2025-10-23T16:31:41-04:00 compilade convert : handle pre-quantized models (#14810)
|
||
0bf47a1dbba4d36f2aff4e8c34b06210ba34e688 dd62dcfab97e420949519fd0eac9fca7bf97e635 2025-10-23T21:30:17+02:00 Johannes Gäßler server: add memory breakdown print (#16740)
|
||
dd62dcfab97e420949519fd0eac9fca7bf97e635 d0660f237a5c31771a3d6d1030ebe3e0c409ba92 2025-10-23T15:54:46+02:00 Julien Denize convert : Make mistral-common dependency optional (#16738)
|
||
d0660f237a5c31771a3d6d1030ebe3e0c409ba92 fe6a9882acf5c02f96624ed8f80144100d7006cb 2025-10-23T15:00:49+02:00 Xuan-Son Nguyen mtmd-cli : allow using --jinja (#16718)
|
||
fe6a9882acf5c02f96624ed8f80144100d7006cb 061f0eff02dc9a82f7bd850db3bd70b8a0b5e87a 2025-10-23T17:07:31+05:30 Prajwal B Mehendarkar Manually link -lbsd to resolve flock symbol on AIX (#16610)
|
||
061f0eff02dc9a82f7bd850db3bd70b8a0b5e87a 8cf6b42d467d05fa7d9776d2bcc69974ecce6900 2025-10-23T19:14:06+08:00 Aman Gupta ggml-cuda: use passed ops instead of hardcoded ops (#16712)
|
||
8cf6b42d467d05fa7d9776d2bcc69974ecce6900 9de9672adb0f4ca4e39483ac3ffed52b3f70a55d 2025-10-23T11:32:24+02:00 matteo server : send partial stop string when <EOG> is reached (#15007)
|
||
9de9672adb0f4ca4e39483ac3ffed52b3f70a55d 63d2fc46e17a06be5b4b5823a5ada088317f1f0a 2025-10-22T20:05:15-05:00 Matthew Michel sycl: use async memory allocation to fix crashes during graph recording (#16644)
|
||
63d2fc46e17a06be5b4b5823a5ada088317f1f0a a2e0088d9242bd9e57f8b852b05a6e47843b5a45 2025-10-22T13:47:09-07:00 Max Krasnyansky Add experimental ggml-hexagon backend for the Hexagon NPU (#16547)
|
||
a2e0088d9242bd9e57f8b852b05a6e47843b5a45 9b9201f65a22c02cee8e300f58f480a588591227 2025-10-22T11:20:55-07:00 Diego Devesa Revert "ggml : Leverage the existing GGML_F32_VEC helpers to vectorize ggml_v…" (#16723)
|
||
9b9201f65a22c02cee8e300f58f480a588591227 19a5a3edfd306516cc419679d69d6435943b6816 2025-10-22T16:58:23+02:00 Pascal webui: introduce OpenAI-compatible model selector in JSON payload (#16562)
|
||
19a5a3edfd306516cc419679d69d6435943b6816 d8eaa26e4d9228df3aa46a930db60c8eaab67c1b 2025-10-22T05:14:14-05:00 sirus20x6 ggml : Leverage the existing GGML_F32_VEC helpers to vectorize ggml_vec_set_f32 for faster fills (#16522)
|
||
d8eaa26e4d9228df3aa46a930db60c8eaab67c1b 9285325ce0174631c3cd6121d56084adc4ef2d8f 2025-10-22T12:01:22+02:00 Acly tests : fix test-thread-safety when compiling with multiple backends (#16699)
|
||
9285325ce0174631c3cd6121d56084adc4ef2d8f 03792ad93609fc67e41041c6347d9aa14e5e0d74 2025-10-22T12:33:08+08:00 Aman Gupta CUDA: fix bug in topk-moe softmax (#16711)
|
||
03792ad93609fc67e41041c6347d9aa14e5e0d74 51d1a8c997bd2629ef211a30208058ea87a30982 2025-10-21T22:40:38+08:00 Aman Gupta CUDA: topk-moe: add optional parameter for gpt-oss (#16649)
|
||
51d1a8c997bd2629ef211a30208058ea87a30982 4926419c4d74a1cf724e7163d937eb72f36e7b26 2025-10-21T15:27:53+02:00 Johannes Gäßler CUDA: better error for FA kernel with 0 occupancy (#16643)
|
||
4926419c4d74a1cf724e7163d937eb72f36e7b26 6ea37f57391d27736c35cd3c20c1f990b7952b74 2025-10-21T16:43:14+08:00 Aman Gupta ggml: add ggml_can_fuse_subgraph (#16662)
|
||
6ea37f57391d27736c35cd3c20c1f990b7952b74 fb349848f387f355450c3187556e71e6d32c145f 2025-10-20T22:26:17-07:00 lhez opencl: fix warnings and clean up profiling (#16688)
|
||
fb349848f387f355450c3187556e71e6d32c145f 6de8ed75196c7cd98c1f34bbf3a7452451ba8ac2 2025-10-20T22:16:08-05:00 Jeff Bolz vulkan: Handle FA with all -inf mask values (#16447)
|
||
6de8ed75196c7cd98c1f34bbf3a7452451ba8ac2 84bf3c677857279037adf67cdcfd89eaa4ca9281 2025-10-21T01:21:12+03:00 YehuditE sycl : add PAD_REFLECT_D1 operator support (#16145)
|
||
84bf3c677857279037adf67cdcfd89eaa4ca9281 c9c1972e2c2cc6a771fcc145bfa138700179f961 2025-10-20T21:38:20+02:00 Sigbjørn Skjæret model : add BailingMoeV2 support (#16063)
|
||
c9c1972e2c2cc6a771fcc145bfa138700179f961 b617cfd2896edd592a36ebbc041817eb030a1005 2025-10-20T19:49:02+02:00 Aleksander Grygier Handle legacy 'context' attachments (#16687)
|
||
b617cfd2896edd592a36ebbc041817eb030a1005 79068501fac9a74cca7129a8e5a8281b410a853e 2025-10-20T05:53:50-07:00 Diego Devesa ggml-alloc : fix leak when reusing a tensor with a larger size (#16679)
|
||
79068501fac9a74cca7129a8e5a8281b410a853e 0e4a0cf2fae667d3efcf52f2f52398779d986b1d 2025-10-20T14:21:12+02:00 Aleksander Grygier Prevent premature submission on IME input (#16673)
|
||
0e4a0cf2fae667d3efcf52f2f52398779d986b1d 13f2cfad4170c096c51a02c24a6a158cb47f1480 2025-10-20T13:29:14+02:00 Aleksander Grygier Import/Export UX improvements (#16619)
|
||
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)
|
||
72d53e6c4decee8b339e49aed8cc0e234b9639dc 2330de7b847ca84eac766df372c604c26db72747 2025-10-20T10:20:04+02:00 Ron Evans readme: update bindings (#16651)
|
||
2330de7b847ca84eac766df372c604c26db72747 7062dd8460685d6700ed7621e50a22c6f3400ca3 2025-10-20T11:08:32+03:00 safranowith SYCL: Add support for FLOOR,CEIL,ROUND and TRUNC unary operators (#16613)
|
||
7062dd8460685d6700ed7621e50a22c6f3400ca3 0398752dd450dfabdd1b9e289f6364c2600f6ab5 2025-10-20T15:44:21+08:00 takuya kodama llama-context: only warn on pooling_type when user specified (#16674)
|
||
c20c0a5c0c44179f5267a262546624854b1203d1 62425b3a7e8b169a3bbe6ce8c7324513a7fcee8c 2025-10-20T14:41:08+08:00 Yunzhe Jia kv: turn on common-prefix mtmd: fix output size bug
|
||
0398752dd450dfabdd1b9e289f6364c2600f6ab5 4f73d0a95120687e2c527739f771330a5271259a 2025-10-19T23:54:31+02:00 Giuseppe Scrivano model : add Granite Hybrid types (#16635)
|
||
4f73d0a95120687e2c527739f771330a5271259a cec5edbcaec69bbf6d5851cabce4ac148be41701 2025-10-20T05:06:39+08:00 Aaron Teo ci : fix binaries release failure for s390x (binaries may not work yet) (#16664)
|
||
cec5edbcaec69bbf6d5851cabce4ac148be41701 fcb235b46618921cbd826acd49b553b5302233aa 2025-10-19T14:03:25+02:00 Sigbjørn Skjæret ci : avoid manual updates of docs/ops.md (#16663)
|
||
fcb235b46618921cbd826acd49b553b5302233aa 55754bebd5d570960cde9c0ba991a0b2991f6b1e 2025-10-19T18:37:47+08:00 Aaron Teo ci: include s390x release binaries (#16648)
|
||
55754bebd5d570960cde9c0ba991a0b2991f6b1e ee09828cb057460b369576410601a3a09279e23c 2025-10-19T15:37:12+08:00 Aman Gupta CODEOWNERS: update for ggml-cuda/mmf (#16660)
|
||
ee09828cb057460b369576410601a3a09279e23c e56abd2098dd2e2b0804691b93c13b48ae421627 2025-10-18T14:47:32+02:00 Johannes Gäßler HIP: fix GPU_TARGETS (#16642)
|
||
e56abd2098dd2e2b0804691b93c13b48ae421627 38355c6c8e43204e11a22daa7483082c0ff01e71 2025-10-18T05:22:57-05:00 Jeff Bolz vulkan: Implement topk_moe fused shader, ported from CUDA (#16641)
|
||
38355c6c8e43204e11a22daa7483082c0ff01e71 81387858f1fbcc1acedbd308486e1016618ca8f8 2025-10-18T17:52:53+08:00 Aman Gupta CUDA: use registers instead of smem in topk-moe (#16647)
|
||
81387858f1fbcc1acedbd308486e1016618ca8f8 66b0dbcb2d462e7b70ba5a69ee8c3899ac2efb1c 2025-10-17T17:55:32-07:00 Shawn Gu opencl: transposed gemm/gemv moe kernel with mxfp4,f32 (#16602)
|
||
66b0dbcb2d462e7b70ba5a69ee8c3899ac2efb1c 41386cf365d894134ee0813d15e2f5d76f6a4d8e 2025-10-17T17:41:09+02:00 Johannes Gäßler llama-model: fix insonsistent ctxs <-> bufs order (#16581)
|
||
41386cf365d894134ee0813d15e2f5d76f6a4d8e 3d4e86bbeb15f487d6da6174ba6191b7c212cc25 2025-10-17T18:02:52+03:00 Radoslav Gerganov rpc : report actual free memory (#16616)
|
||
3d4e86bbeb15f487d6da6174ba6191b7c212cc25 342c728d031d50673feded797520a44127d73379 2025-10-17T14:23:47+02:00 Giuseppe Scrivano vulkan: Add State Space Model (SSM) Operations Support (#16463)
|
||
342c728d031d50673feded797520a44127d73379 ababae7e1ec3e9cfdab0322ee55ea3389e82a4d5 2025-10-17T18:01:23+08:00 muggle-stack ggml : fix SpaceMit IME array out-of-bounds in task assignment (#16629)
|
||
ababae7e1ec3e9cfdab0322ee55ea3389e82a4d5 b19491599d4d42c606601d75e95c1d1de3291f8e 2025-10-17T10:35:03+02:00 Pascal webui: reorganize settings layout (#16607)
|
||
b19491599d4d42c606601d75e95c1d1de3291f8e 9ad4f1931ee0f3b41d9355245ef744786aaae0aa 2025-10-17T02:31:04-05:00 Jeff Bolz vulkan: fix debug build (add_rms_len/data not found) (#16624)
|
||
9ad4f1931ee0f3b41d9355245ef744786aaae0aa 79967ec596c0dacfd2251b085a57e79df292b1cc 2025-10-17T02:33:58-04:00 Ilia Ilmer metal : add `CONV_TRANSPOSE_2D` (#16542)
|
||
79967ec596c0dacfd2251b085a57e79df292b1cc ceff6bb253dd306f5404d7ccb3f11fadafe71b52 2025-10-17T06:59:31+01:00 Olivier Chafik grammar : use int64_t to avoid int overflows in int schema to grammar conversion logic (#16626)
|
||
ceff6bb253dd306f5404d7ccb3f11fadafe71b52 1bb4f43380944e94c9a86e305789ba103f5e62bd 2025-10-17T05:36:40+03:00 GittyBurstein SYCL SET operator optimized for F32 tensors (#16350)
|
||
1bb4f43380944e94c9a86e305789ba103f5e62bd 683fa6ba4ed3a23b939d3e11e6dc860bd47a0ccf 2025-10-16T19:00:31+02:00 Xuan-Son Nguyen mtmd : support home-cooked Mistral Small Omni (#14928)
|
||
683fa6ba4ed3a23b939d3e11e6dc860bd47a0ccf b22572e97dc51757d3ebe917a5a283385010ec68 2025-10-16T16:28:41+02:00 Pascal fix: added a normalization step for MathJax-style \[\] and \(\) delimiters (#16599)
|
||
b22572e97dc51757d3ebe917a5a283385010ec68 7a50cf388a530127cbbdd2a507ef81a451c9d819 2025-10-16T16:26:21+03:00 GittyBurstein sycl : add ARANGE operator (#16362)
|
||
62425b3a7e8b169a3bbe6ce8c7324513a7fcee8c 76a63a2998286c1649de2496bccb7d8a46549895 2025-10-16T17:34:39+08:00 Yunzhe Jia kv: let seq_rm free whole seq since current hw-op cannot support common-prefix
|
||
7a50cf388a530127cbbdd2a507ef81a451c9d819 6f5d924637a15abedb111cbbffd7da5f31c81855 2025-10-16T16:41:11+08:00 Chenguang Li CANN: format code using .clang-format (#15863)
|
||
76a63a2998286c1649de2496bccb7d8a46549895 b67217078aa748b06e1b2269d88dd18c0aca2c26 2025-10-16T16:31:15+08:00 Yunzhe Jia kv: fix common-prefix bug by implementing llama_memory_seq_rm
|
||
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)
|
||
ee50ee1eadff58777ae746827b04de7ba0befc55 7adc79c03234de9a20661fd6dbf2d02c32ca7acb 2025-10-16T07:21:28+03:00 yael-works SYCL: Add GGML_OP_MEAN operator support (#16009)
|
||
7adc79c03234de9a20661fd6dbf2d02c32ca7acb 466c1911ab736f0b7366127edee99f8ee5687417 2025-10-15T22:43:08+02:00 Aleksei Nikiforov gguf-py : add support for endian conversion of BF16 data (#16594)
|
||
466c1911ab736f0b7366127edee99f8ee5687417 0cb7a0683b0529172472d74d21f05470a607f297 2025-10-15T22:24:51+03:00 safranowith cpu : add FLOOR, CEIL, ROUND and TRUNC unary operators (#16083)
|
||
0cb7a0683b0529172472d74d21f05470a607f297 d93f8439b08c4f35e13a41a7366901fdbe770fc8 2025-10-15T10:51:04-07:00 lhez opencl: add q8_0 mm support (#16469)
|
||
d93f8439b08c4f35e13a41a7366901fdbe770fc8 f9fb33f2630b4b4ba9081ce9c0c921f8cd8ba4eb 2025-10-15T10:48:28-07:00 lhez opencl: fix FA for f32 (#16584)
|
||
f9fb33f2630b4b4ba9081ce9c0c921f8cd8ba4eb f4ce81c45e7bd910e36bf44c253fc5255c49b1e4 2025-10-15T16:22:20+02:00 Aleksander Grygier Add server-driven parameter defaults and syncing (#16515)
|
||
f4ce81c45e7bd910e36bf44c253fc5255c49b1e4 17304cbcc1dd24de7741cbe57925d58e90a98ac1 2025-10-15T23:05:56+09:00 Sam/Samuel metal: optimise `GGML_OP_SUM` (#16559)
|
||
17304cbcc1dd24de7741cbe57925d58e90a98ac1 3e3cb19f6449f9168a128eaeae01f8f41b049acc 2025-10-15T16:53:12+03:00 Georgi Gerganov server : fix img token logs (#16595)
|
||
3e3cb19f6449f9168a128eaeae01f8f41b049acc 5acd455460f457942d8dd02e3dd9b1eebfce99fe 2025-10-15T14:48:08+02:00 Xuan-Son Nguyen llama-quant: add support for mmproj (#16592)
|
||
5acd455460f457942d8dd02e3dd9b1eebfce99fe 554fd578a5ed78a10f371f5850c6a69aa83df15a 2025-10-15T13:54:15+02:00 Julius Tischbein CUDA: Changing the CUDA scheduling strategy to spin (#16585)
|
||
554fd578a5ed78a10f371f5850c6a69aa83df15a fa882fd2b1bcb663de23af06fdc391489d05b007 2025-10-15T12:51:27+03:00 Georgi Gerganov server : fix mtmd checkpoints (#16591)
|
||
fa882fd2b1bcb663de23af06fdc391489d05b007 ffa059034c1e41f3e58363ef3c46d2fcf854bc4d 2025-10-14T20:33:05+03:00 Georgi Gerganov metal : avoid using Metal's gpuAddress property (#16576)
|
||
ffa059034c1e41f3e58363ef3c46d2fcf854bc4d 120bf7046d85a893f064c12abe58bfeebd735f84 2025-10-14T19:18:05+02:00 SavicStefan vulkan: Add ACC_TYPE_VEC2 implementation (#16203)
|
||
120bf7046d85a893f064c12abe58bfeebd735f84 4258e0cfe72c72ad2787be1e9f93253e89219455 2025-10-14T22:48:08+08:00 Aman Gupta CUDA + openCL: fix bug in accessing rms_norm->src while doing fusion (#16577)
|
||
4258e0cfe72c72ad2787be1e9f93253e89219455 7ea15bb64c81e3813eb0babf9a57e1bc5697f569 2025-10-14T08:53:37-05:00 Jeff Bolz vulkan: Support FA with K/V in F32 (#16543)
|
||
7ea15bb64c81e3813eb0babf9a57e1bc5697f569 9c7185dd28416cf67f5e3b268381f311b5e3da56 2025-10-14T07:51:36-05:00 Jeff Bolz vulkan: Improve build time for MSVC (#16545)
|
||
9c7185dd28416cf67f5e3b268381f311b5e3da56 1ee9d0b415cdf5240418c110a18b419f4002b154 2025-10-14T14:22:47+02:00 Johannes Gäßler CUDA: enable FA for FP32 KV cache (#16546)
|
||
1ee9d0b415cdf5240418c110a18b419f4002b154 48e2fa9fb7c2de1e53808fdb65ec33f916020fc4 2025-10-14T19:16:21+08:00 Aman Gupta CUDA: use fastdiv + ggml_cuda_mad for mmvf (#16557)
|
||
48e2fa9fb7c2de1e53808fdb65ec33f916020fc4 5b6913c47b6bc71a6f927805a45387d5657d8b89 2025-10-14T19:15:15+08:00 Aman Gupta CUDA: add fp kernel for larger batch size MoE (#16512)
|
||
5b6913c47b6bc71a6f927805a45387d5657d8b89 bc07349a7f87ba6eb31ed4b0ea9d9a7352185213 2025-10-14T09:53:49Z Anav Prasad cuda : remove legacy copy-op pointer indirection code (#16485)
|
||
bc07349a7f87ba6eb31ed4b0ea9d9a7352185213 e60f241eacec42d3bd7c9edd37d236ebf35132a8 2025-10-14T08:48:50+03:00 Georgi Gerganov server : dynamic token limit for prompt cache (#16560)
|
||
e60f241eacec42d3bd7c9edd37d236ebf35132a8 e38b7c6e9e4453e3b3e96d76e38bc2ccb6bce458 2025-10-13T23:07:57+03:00 Georgi Gerganov metal : FA support F32 K and V and head size = 32 (#16531)
|
||
e38b7c6e9e4453e3b3e96d76e38bc2ccb6bce458 5016b7286240d29f8f640039989b84ea3a854344 2025-10-13T22:42:37+03:00 Georgi Gerganov graph : support cacheless embeddings with FA and iSWA (#16528)
|
||
5016b7286240d29f8f640039989b84ea3a854344 7049736b2dd9011bf819e298b844ebbc4b5afdc9 2025-10-13T11:50:37-07:00 lhez opencl: fix build targeting CL 2 (#16554)
|
||
7049736b2dd9011bf819e298b844ebbc4b5afdc9 01d2bdc2bc61b676706830305b286b08b9885a41 2025-10-13T16:29:45+02:00 Johannes Gäßler CUDA: fix numerical issues in tile FA kernel (#16540)
|
||
01d2bdc2bc61b676706830305b286b08b9885a41 56fc38b9655fbe1869d8bd6cfb269418196cea69 2025-10-13T20:48:47+08:00 Jie Fu (傅杰) ggml : fix build broken with -march=armv9-a on MacOS (#16520)
|
||
56fc38b9655fbe1869d8bd6cfb269418196cea69 1fb9504eb744969a990bfe4cfcf1d3d7a479541c 2025-10-13T17:01:24+08:00 Chenguang Li CANN: fix CPU memory leak in CANN backend (#16549)
|
||
1fb9504eb744969a990bfe4cfcf1d3d7a479541c 3f750f8d760ab5a61491e6a9409072dfeee4b4d7 2025-10-13T10:55:32+02:00 Pascal fix: add remark plugin to render raw HTML as literal text (#16505)
|
||
b67217078aa748b06e1b2269d88dd18c0aca2c26 98c8269abefdfa8fde47dfb2769ceebc8b704e86 2025-10-13T11:48:14+08:00 Yunzhe Jia add different path for unified and seperate compilation
|
||
3f750f8d760ab5a61491e6a9409072dfeee4b4d7 c515fc577166042234241c6bd0da9b08dcbe2bb9 2025-10-13T16:25:02+08:00 Sam/Samuel metal: add support for opt_step_sgd (#16539)
|
||
c515fc577166042234241c6bd0da9b08dcbe2bb9 f9bc66c3ebcfddb5f09e4b21253623caeb8e414a 2025-10-13T11:22:27+03:00 Georgi Gerganov ggml : fix scalar path for computing norm (#16558)
|
||
f9bc66c3ebcfddb5f09e4b21253623caeb8e414a a31cf36ad946a13b3a646bf0dadf2a481e89f944 2025-10-13T08:52:22+08:00 hipudding CANN: Update several operators to support FP16 data format (#16251)
|
||
a31cf36ad946a13b3a646bf0dadf2a481e89f944 81d54bbfd599811b354c39f04550888168be7780 2025-10-13T02:43:14+08:00 Sam/Samuel metal : add opt_step_adamw and op_sum (#16529)
|
||
81d54bbfd599811b354c39f04550888168be7780 c7be9febcbafa9af7d1b9443f86475c59c9c5f87 2025-10-12T18:06:41+02:00 Pascal webui: remove client-side context pre-check and rely on backend for limits (#16506)
|
||
c7be9febcbafa9af7d1b9443f86475c59c9c5f87 8415f61e23d04427cd0d912fbb9d33b85f849456 2025-10-12T21:53:35+08:00 Neo Zhang Jianyu [SYCL] fix UT fault cases: count-equal, argsort, pad OPs (#16521)
|
||
8415f61e23d04427cd0d912fbb9d33b85f849456 2c301e91abb92d03c1a682b4b540ba835562a74b 2025-10-12T15:48:03+02:00 Mathieu Baudier ci : add Vulkan on Ubuntu with default packages build (#16532)
|
||
2c301e91abb92d03c1a682b4b540ba835562a74b 4b2dae383df708e2afc49c4859a81cd074f5ac10 2025-10-12T08:18:47-05:00 Aldehir Rojas common : handle unicode during partial json parsing (#16526)
|
||
4b2dae383df708e2afc49c4859a81cd074f5ac10 41aac5c69b5fb281bc1f486afb053f78101bb39e 2025-10-12T09:29:13+03:00 Georgi Gerganov common : update presets (#16504)
|
||
41aac5c69b5fb281bc1f486afb053f78101bb39e a2fba89a426ff8005d303c73f0436e7e67368b70 2025-10-12T00:25:37-05:00 sirus20x6 ggml : Fix FP16 ELU positive branch (#16519)
|
||
a2fba89a426ff8005d303c73f0436e7e67368b70 20cc625edc2264aae2779e71bef1593e6a4e8c43 2025-10-12T07:19:06+02:00 Daniel Bevenius hparams : add check for layer index in is_recurrent (#16511)
|
||
20cc625edc2264aae2779e71bef1593e6a4e8c43 11f0af5504252e453d57406a935480c909e3ff37 2025-10-12T00:15:00-05:00 sirus20x6 ggml: Correct SVE implementation in ggml_vec_dot_f16_unroll (#16518)
|
||
11f0af5504252e453d57406a935480c909e3ff37 a3cb04744fb5c591985f53b749fef5407d07a145 2025-10-11T20:54:32+02:00 Johannes Gäßler CUDA: faster tile FA, add oob checks, more HSs (#16492)
|
||
a3cb04744fb5c591985f53b749fef5407d07a145 4a8fbe0a5eb75f339782ebcf29c17848122184d3 2025-10-11T16:54:10+03:00 Georgi Gerganov metal : fix mul-mm condition + fix mul-mv permuted kernels (#16494)
|
||
4a8fbe0a5eb75f339782ebcf29c17848122184d3 31d0ff1869aa2ea31f4e96d5877d0343e9a2171b 2025-10-11T15:50:49+02:00 Pascal feat: render user content as markdown option (#16358)
|
||
31d0ff1869aa2ea31f4e96d5877d0343e9a2171b 97870e64975b26c5e06a3540a8dc0ff601351e86 2025-10-11T21:39:04+08:00 Yann Follet server / ranking : add sorting and management of top_n (#16403)
|
||
97870e64975b26c5e06a3540a8dc0ff601351e86 477a66b03501cf3bd067f8968b77ca4d053ff1bd 2025-10-11T04:02:26-07:00 Diego Devesa cuda : avoid initializing unused devices (#16510)
|
||
477a66b03501cf3bd067f8968b77ca4d053ff1bd e60f01d941bc5b7fae62dd57fee4cec76ec0ea6e 2025-10-11T11:33:41+03:00 amirai21 convert : correctly handle LLaMA tokenizer for Jamba (#16470)
|
||
e60f01d941bc5b7fae62dd57fee4cec76ec0ea6e 81086cd6a3ca1252f0dc0f938171648399179c53 2025-10-10T22:15:05+03:00 Georgi Gerganov server : fix division by zero when reporting stats (#16501)
|
||
81086cd6a3ca1252f0dc0f938171648399179c53 68ee98ae181a5c83a5cc6261daeee69a1f588c15 2025-10-10T17:17:31+03:00 Georgi Gerganov vocab : mark EOT token for Granite models (#16499)
|
||
68ee98ae181a5c83a5cc6261daeee69a1f588c15 cdb6da468cc33323955a523738d2e1675aeb5e9a 2025-10-10T17:11:07+03:00 Radoslav Gerganov server : return HTTP 400 if prompt exceeds context length (#16486)
|
||
cdb6da468cc33323955a523738d2e1675aeb5e9a 6d69ab3f262eca3f0c6ad0ce075d2c80b9924d0e 2025-10-10T13:22:27+03:00 Radoslav Gerganov server : log requests to /v1/completions (#16495)
|
||
6d69ab3f262eca3f0c6ad0ce075d2c80b9924d0e 1faa13a1187051af66b0fd9f0d6effe4c77f0b3e 2025-10-10T13:45:46+05:30 Prajwal B Mehendarkar cmake : Dont define XOPENSOURCE on AIX (#16481)
|
||
1faa13a1187051af66b0fd9f0d6effe4c77f0b3e 1deee0f8d494981c32597dca8b5f8696d399b0f2 2025-10-09T22:54:57+02:00 Pascal webui: updated the chat service to only include max_tokens in the req… (#16489)
|
||
1deee0f8d494981c32597dca8b5f8696d399b0f2 d00cbea63c671cd85a57adaa50abf60b3b87d86f 2025-10-09T22:11:15+03:00 duduta cpu : optimize the ggml NORM operation (#15953)
|
||
d00cbea63c671cd85a57adaa50abf60b3b87d86f 8328fd4bae76fc027f8ca0e9a05acd3788dabe3f 2025-10-09T18:54:51+03:00 Georgi Gerganov server : host-memory prompt caching (#16391)
|
||
8328fd4bae76fc027f8ca0e9a05acd3788dabe3f 56b4795842d852152222ca7a2d4304008facf1b9 2025-10-09T17:36:29+02:00 Pascal No markdown in cot (#16483)
|
||
56b4795842d852152222ca7a2d4304008facf1b9 2c0d875ae6c6043fbbafd2831e160955e9ca6af1 2025-10-09T14:35:22+02:00 Daniel Bevenius model-conversion : add support for SentenceTransformers (#16387)
|
||
2c0d875ae6c6043fbbafd2831e160955e9ca6af1 aa4711d369b0d4adb6802b98ad6ea362f838710e 2025-10-09T09:13:18+01:00 sudhiarm ci: add ARM64 Kleidiai build and test support (#16462)
|
||
aa4711d369b0d4adb6802b98ad6ea362f838710e d80d6d2400b8faa80654c723cb5bdf9fc8f4db06 2025-10-09T15:50:25+08:00 Chenguang Li CANN: Improve ACL graph matching (#16166)
|
||
d80d6d2400b8faa80654c723cb5bdf9fc8f4db06 b2602137557b2b28a39e03612717d85ead9a6f5a 2025-10-09T09:29:17+02:00 Charles Xu kleidiai: kernel interface refactoring (#16460)
|
||
b2602137557b2b28a39e03612717d85ead9a6f5a e08db4259521de493b7aeb49dadf29ebd1ee966a 2025-10-09T15:25:11+08:00 Neo Zhang Jianyu [SYCL] refactor soft_max, add soft_max_back (#16472)
|
||
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)
|
||
9d0882840e6c3fb62965d03af0e22880ea90e012 d2ee056e1df0fdf646270fb5621e9f92084b59a7 2025-10-08T20:21:46+02:00 ai-fonsi Disable CUDA host buffers on integrated GPUs (#16308)
|
||
d2ee056e1df0fdf646270fb5621e9f92084b59a7 b2c08c9ec4cd59ed88c1ebdd94f109af5ccc978e 2025-10-08T17:20:18+09:00 issixx server : fix cancel pending task (#16467)
|
||
b2c08c9ec4cd59ed88c1ebdd94f109af5ccc978e 7fdd16b432d247121fe5fe7b21f8805f85266c85 2025-10-08T10:57:53+03:00 Georgi Gerganov metal : mark FA blocks (#16372)
|
||
7fdd16b432d247121fe5fe7b21f8805f85266c85 74b8fc17f92ada295a648e3c5eb28f46bca7d892 2025-10-08T10:57:29+03:00 Georgi Gerganov server : improve context checkpoint logic (#16440)
|
||
74b8fc17f92ada295a648e3c5eb28f46bca7d892 aeaf8a36f06b5810f5ae4bbefe26edb33925cf5e 2025-10-07T13:48:56-07:00 Reese Levine ggml webgpu: profiling, CI updates, reworking of command submission (#16452)
|
||
aeaf8a36f06b5810f5ae4bbefe26edb33925cf5e df1b612e29ba97a2e67db339b1e8c7465702b7e8 2025-10-07T20:03:35+02:00 Tarek Dakhran llama : support LiquidAI LFM2-MoE hybrid model (#16464)
|
||
df1b612e29ba97a2e67db339b1e8c7465702b7e8 4e0388aa8a1f4ef1065701dc9c3947aea0a1b9a5 2025-10-07T15:57:14+03:00 Georgi Gerganov server : add `/v1/health` endpoint (#16461)
|
||
4e0388aa8a1f4ef1065701dc9c3947aea0a1b9a5 ef4c5b87ea2556ff8ca99cca3abdf48bdbca22f2 2025-10-07T11:11:08+02:00 Sascha Rogmann webui : added download action (#13552) (#16282)
|
||
ef4c5b87ea2556ff8ca99cca3abdf48bdbca22f2 c61ae20d05bd4fdd8551311325a2845336449426 2025-10-07T10:32:32+03:00 Georgi Gerganov presets : fix pooling param for embedding models (#16455)
|
||
c61ae20d05bd4fdd8551311325a2845336449426 0123ff38f53d34752f29239a29d0e40a6dc4110f 2025-10-07T09:59:13+03:00 Radoslav Gerganov rpc : update documentation (#16441)
|
||
0123ff38f53d34752f29239a29d0e40a6dc4110f 0a319bb75ed29d968e2a9b544011b09ccb932915 2025-10-07T08:24:17+03:00 Georgi Gerganov memory : use sequential equal splits for recurrent modules (#16442)
|
||
0a319bb75ed29d968e2a9b544011b09ccb932915 1d6092fc72f4d10f4486ac95edfd414bc08b62b8 2025-10-07T08:23:30+03:00 Georgi Gerganov metal : add support for non-padded FA KV (#16148)
|
||
1d6092fc72f4d10f4486ac95edfd414bc08b62b8 8ae32dc9ecd7aeebf3a5b43557e0552a0a04cd4f 2025-10-07T08:22:35+03:00 Georgi Gerganov tests : add -INF blocks to the KQ mask in the FA tests (#16380)
|
||
8ae32dc9ecd7aeebf3a5b43557e0552a0a04cd4f 3df2244df40c67dfd6ad548b40ccc507a066af2b 2025-10-07T08:21:40+03:00 Georgi Gerganov metal : various optimizations + refactoring (#16446)
|
||
3df2244df40c67dfd6ad548b40ccc507a066af2b c08002a1988348403a5fd59b1fa3de3a10a6f92f 2025-10-06T12:55:53-05:00 Gadflyii llama : add --no-host to disable host buffers (#16310)
|
||
c08002a1988348403a5fd59b1fa3de3a10a6f92f 3a002afafa8e06555f11aed88b02d055d8a166f3 2025-10-06T10:59:40-06:00 Gabe Goodhart chat : Granite Docling stopping (#16438)
|
||
3a002afafa8e06555f11aed88b02d055d8a166f3 a23b9bdbd3b64ce172f9962249f432d01aea7437 2025-10-06T17:40:21+02:00 Sigbjørn Skjæret ci : refactor sdk caching to minimize storage (#16414)
|
||
a23b9bdbd3b64ce172f9962249f432d01aea7437 04e632a4aab8e6bfff0f8bc216b36ceb1e199ff9 2025-10-06T16:05:27+03:00 Georgi Gerganov ggml : fix unaligned access in AMX code (#16315)
|
||
04e632a4aab8e6bfff0f8bc216b36ceb1e199ff9 a80ff183abe4e5a76316257ffa597da41b3b6fa0 2025-10-06T14:56:59+02:00 Daniel Bevenius ci : remove missing reranker model files (#16444)
|
||
a80ff183abe4e5a76316257ffa597da41b3b6fa0 1d49ca37594fb49db6aa9518ba7c512e5ccd0108 2025-10-06T14:17:12+02:00 Daniel Bevenius ggml-cpu : fix leftover handling in ggml_vec_scale_f32 for SVE (#16443)
|
||
1d49ca37594fb49db6aa9518ba7c512e5ccd0108 c5fef0fcea3b978a2318b1af170209ecec7c37b4 2025-10-06T09:29:56Z Yuannan nix : removed metal for nix (#16118)
|
||
c5fef0fcea3b978a2318b1af170209ecec7c37b4 ca71fb9b368e3db96e028f80c4c9df6b6b370edd 2025-10-06T03:53:31-04:00 Oleksandr Kuvshynov server: update readme to mention n_past_max metric (#16436)
|
||
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)
|
||
f39283960b58a92ecc0c72567711318b20e22b55 898acba6816ad23b6a9491347d30e7570bffadfd 2025-10-04T16:22:45+03:00 Radoslav Gerganov rpc : check src buffer when copying tensor (#16421)
|
||
898acba6816ad23b6a9491347d30e7570bffadfd e29acf74fea996014380d59d31aa504ae8964258 2025-10-04T12:49:16+03:00 Radoslav Gerganov rpc : add support for multiple devices (#16276)
|
||
e29acf74fea996014380d59d31aa504ae8964258 128d522c04286e019666bd6ee4d18e3fbf8772e2 2025-10-04T11:42:56+02:00 Acly vulkan : incremental shader builds (#16341)
|
||
128d522c04286e019666bd6ee4d18e3fbf8772e2 f6dcda390004b627ef30af378d0c01ad2519289e 2025-10-03T20:51:48+02:00 Pascal chat : support Magistral thinking (#16413)
|
||
f6dcda390004b627ef30af378d0c01ad2519289e 606a73f53175077429484b23dcf799f69a31d0bd 2025-10-03T13:34:51-05:00 ddh0 server : context checkpointing for hybrid and recurrent models (#16382)
|
||
606a73f53175077429484b23dcf799f69a31d0bd 946f71ed9ade07e319859b5ce656144140e066fb 2025-10-03T19:18:56+03:00 Georgi Gerganov metal : fix loop bound in ggml_mem_ranges (#16412)
|
||
946f71ed9ade07e319859b5ce656144140e066fb 638d330246954e88dffc22ce01fec15e6894e544 2025-10-03T14:40:25+02:00 Sigbjørn Skjæret llama : fix shapes for bert/mpt q/k norm (#16409)
|
||
638d330246954e88dffc22ce01fec15e6894e544 84c8e305e8010a1a3d43bdd0a25f737ac67809a4 2025-10-03T13:49:08+02:00 Acly ggml : fix graph reallocation with multiple chunks (#16396)
|
||
84c8e305e8010a1a3d43bdd0a25f737ac67809a4 2aaf0a2a2056d75d0dd53ab8a181473760e6ab22 2025-10-03T12:51:40+02:00 Aleksander Grygier Fix missing messages on sibling navigation (#16408)
|
||
2aaf0a2a2056d75d0dd53ab8a181473760e6ab22 0e1f83855609d73beaf05d818640b6cfd39d287b 2025-10-03T05:50:46-05:00 Jeff Bolz vulkan: Replace uses of maxMemoryAllocationSize and VK_WHOLE_SIZE (#16354)
|
||
0e1f83855609d73beaf05d818640b6cfd39d287b ad126479c25cf983a0f994a08ba0911cf49ed62b 2025-10-03T04:52:46-05:00 Jeff Bolz vulkan: Fix FA coopmat1 invalid array indexing (#16365)
|
||
ad126479c25cf983a0f994a08ba0911cf49ed62b 77233277c912d495d1069543ca540c21a2f9e5b4 2025-10-03T11:45:16+02:00 Daniel Bevenius ci : change macos-13 to macos-15-intel (#16401)
|
||
77233277c912d495d1069543ca540c21a2f9e5b4 e308efda8e54e3f07578937a45a5d83624eb7970 2025-10-03T11:30:39+02:00 Aleksander Grygier Capture model name only after first token (streaming) or completed request (#16405)
|
||
e308efda8e54e3f07578937a45a5d83624eb7970 136bda78c5679b266362b39c58338fff46fd2592 2025-10-03T03:33:08-05:00 Jeff Bolz vulkan: in flash attention, bounds check against nem1 (don't rely on GGML_KQ_MASK_PAD) (#16316)
|
||
136bda78c5679b266362b39c58338fff46fd2592 5113efd34ceda709292de26c72716c49e024fb32 2025-10-03T09:11:34+02:00 Aleksander Grygier webui : Fix messages payload sent to chat completions (#16402)
|
||
5113efd34ceda709292de26c72716c49e024fb32 d64c8104f090b27b1f99e8da5995ffcfa6b726e2 2025-10-03T08:01:31+02:00 Pascal fix: track viewportHeight via window.innerHeight to avoid unwanted scrolling (#16356)
|
||
d64c8104f090b27b1f99e8da5995ffcfa6b726e2 ef07a4090672a3438d7f64f197795d7dc1c18957 2025-10-02T20:10:12+02:00 Sigbjørn Skjæret test-barrier : do not use more threads than physically available (#16389)
|
||
ef07a4090672a3438d7f64f197795d7dc1c18957 34fcc5a4ace8c69476ef2ea3857f39a60334acc4 2025-10-02T11:00:31-07:00 Reese Levine ggml webgpu: add support for soft_max, optimize rms_norm (#16357)
|
||
34fcc5a4ace8c69476ef2ea3857f39a60334acc4 91a2a5655658bb9ab77894716b82fae7ecb4b4d1 2025-10-02T19:43:22+02:00 Piotr Wilkin (ilintar) model : Apertus model implementation (#15852)
|
||
91a2a5655658bb9ab77894716b82fae7ecb4b4d1 72ee736c447be4ededb51ab97904b4d33c90c261 2025-10-02T21:29:56+08:00 R0CKSTAR musa: update compile flags (#16265)
|
||
72ee736c447be4ededb51ab97904b4d33c90c261 f09aefaa84d7f4d5df3f400f67944b94fef5b795 2025-10-02T13:51:36+02:00 Sigbjørn Skjæret ci : fix ubuntu-latest-cmake-rpc (disable ccache) (#16388)
|
||
f09aefaa84d7f4d5df3f400f67944b94fef5b795 bbd32bc0384fbfcf07369617de58856b1e0e95a3 2025-10-02T08:10:07Z Eve ci: update vulkan ci (#16294)
|
||
bbd32bc0384fbfcf07369617de58856b1e0e95a3 2be72c2b121ee99f33927149265ce6073ade9e59 2025-10-02T10:35:43+03:00 Georgi Gerganov ci : fix clean-up of old logs (#16381)
|
||
2be72c2b121ee99f33927149265ce6073ade9e59 95ce0985449c52fb9d6849b95563f7cf933ea0e3 2025-10-02T15:16:25+08:00 Neo Zhang Jianyu SYCL: Update to oneAPI 2025.2 (#16371)
|
||
95ce0985449c52fb9d6849b95563f7cf933ea0e3 c8dedc9999eccf7821a9fe5b29f10e8d075e2217 2025-10-02T05:52:59+02:00 uvos HIP: add IMbackK to codeowner (#16375)
|
||
c8dedc9999eccf7821a9fe5b29f10e8d075e2217 e95fec640f43623911a2cd5bda8b19b1898c530c 2025-10-01T23:32:39+02:00 uvos CI: reenable cdna in rocm docker builds (#16376)
|
||
e95fec640f43623911a2cd5bda8b19b1898c530c ded67b94446ef4f7fd988dbde7a12deef9870c13 2025-10-01T23:09:25+02:00 uvos HIP: Disable ROCWMMA fattn on CDNA when compiled against ROCWMMA 2.0.0 (#16221)
|
||
ded67b94446ef4f7fd988dbde7a12deef9870c13 1fe4e38cc20af058ed320bd46cac934991190056 2025-10-02T06:08:15+09:00 Shunta Saito llama : parameter conversion and loading fixes for PLaMo2 variants (#16075)
|
||
1fe4e38cc20af058ed320bd46cac934991190056 4201deae9c2ae4732db8957b6ce0808d02ec597c 2025-10-01T20:18:03+02:00 uvos ci: Properly install rocwmma for hip builds (#16305)
|
||
4201deae9c2ae4732db8957b6ce0808d02ec597c 764799279f801c696503bacca490b4358587d94c 2025-10-01T19:22:18+02:00 Adrien Gallouët common: introduce http.h for httplib-based client (#16373)
|
||
764799279f801c696503bacca490b4358587d94c 2a9b63383a448ec18c754dfdc6e95cb853940a52 2025-10-01T18:18:10+02:00 Aleksander Grygier Conversation action dialogs as singletons from Chat Sidebar + apply conditional rendering for Actions Dropdown for Chat Conversation Items (#16369)
|
||
2a9b63383a448ec18c754dfdc6e95cb853940a52 1104ca1a1c926b832274694a2773a17066982ad0 2025-10-01T15:54:42+02:00 Aleksander Grygier Improve code block color theming (#16325)
|
||
1104ca1a1c926b832274694a2773a17066982ad0 4f1575921cac9b489fe8f8bbf20aff985e7da45e 2025-10-01T14:09:52+02:00 Sigbjørn Skjæret ci : use registry cache for docker builds (#16366)
|
||
4f1575921cac9b489fe8f8bbf20aff985e7da45e 132d673554e65282e92505f4f77401ab971528bb 2025-10-01T12:08:16+02:00 Aleksander Grygier Add optional setting for showing "Model used:" information (#16337)
|
||
132d673554e65282e92505f4f77401ab971528bb aa9538a63aef35f0224b2502f13b19d650a8ce04 2025-10-01T07:56:36Z Eve vulkan: make ggml_vk_default_dispatcher support older vulkan headers (#16345)
|
||
aa9538a63aef35f0224b2502f13b19d650a8ce04 e74c92e84236b2bab3f3c77bee4ead94928be360 2025-10-01T07:40:26+02:00 Aleksander Grygier webui: Remove running `llama-server` within WebUI `dev.sh` script (#16363)
|
||
e74c92e84236b2bab3f3c77bee4ead94928be360 b2ba81dbe07b6dbea9c96b13346c66973dede32c 2025-09-30T16:24:36-04:00 Bartowski model : support GLM 4.6 (make a few NextN/MTP tensors not required) (#16359)
|
||
b2ba81dbe07b6dbea9c96b13346c66973dede32c bf6f3b3a1965d70e07ca94aab7b01268fe483e96 2025-09-30T21:41:42+02:00 Sigbjørn Skjæret ci : fix ccache key for ubuntu-cpu-cmake (#16355)
|
||
bf6f3b3a1965d70e07ca94aab7b01268fe483e96 7c156df4148eb524b22f7f363bfd2df00f264e0b 2025-09-30T19:52:41+02:00 Adrien Gallouët common : disable progress bar without a tty (#16352)
|
||
7c156df4148eb524b22f7f363bfd2df00f264e0b 16b0ca0d2e63fc1a0a43795f8552f9cb61d9f7a5 2025-09-30T10:45:45-07:00 lhez opencl: support pad_ext (#15888)
|
||
16b0ca0d2e63fc1a0a43795f8552f9cb61d9f7a5 8d78cd2613ccdeb3cc86f59bc8f9ddd31cfbd3ed 2025-09-30T19:18:54+02:00 Pascal Chatapi ignore empty sampling (#16330)
|
||
8d78cd2613ccdeb3cc86f59bc8f9ddd31cfbd3ed d1c84a662daa91be975863913a59975b11458141 2025-09-30T09:57:51-07:00 Reese Levine ggml webgpu: support for rope,div,sub,glu,scale,cont operators (#16187)
|
||
d1c84a662daa91be975863913a59975b11458141 364a7a6d4a786e98947c8a90430ea581213c0ba9 2025-09-30T09:55:13-07:00 lhez opencl: support ne3 in get_rows (#15866)
|
||
364a7a6d4a786e98947c8a90430ea581213c0ba9 2df5bcf357dba0c49b4df1d684b2ffc9e88b7054 2025-09-30T16:39:44+02:00 Adrien Gallouët common : remove common_has_curl() (#16351)
|
||
2df5bcf357dba0c49b4df1d684b2ffc9e88b7054 075c01567bb7e93965ae60b7e119182c3e68f3e9 2025-09-30T15:38:01+02:00 Sigbjørn Skjæret ci : disable ccache for android (#16348)
|
||
075c01567bb7e93965ae60b7e119182c3e68f3e9 a014310374a16f9204f2bcc1b458fc1eda67e469 2025-09-30T13:42:39+03:00 Georgi Gerganov ggml : bump version to 0.9.4 (ggml/1363)
|
||
98c8269abefdfa8fde47dfb2769ceebc8b704e86 2790ecc304e384d9075c55d0628d8bf440025dc4 2025-09-30T17:42:57+08:00 Yunzhe Jia server: adapt to calrt
|
||
2790ecc304e384d9075c55d0628d8bf440025dc4 f528e97d894b4d93934a0711a00253eea980e799 2025-09-30T17:37:58+08:00 Yunzhe Jia adapt to calrt
|
||
f528e97d894b4d93934a0711a00253eea980e799 dcbe998a6d8f13b6c9cb36bbe7b355ced2ac3842 2025-09-11T14:13:35+08:00 Yunzhe Jia add calrt mtmd support
|
||
a014310374a16f9204f2bcc1b458fc1eda67e469 35fb82497ec6c5904b0adb7e1c881a76c1c692db 2025-09-30T08:13:22Z anavp-nvidia cuda : Enable CUDA Graph usage for Nemotron Nano v2 (NemotronH) (#16328)
|
||
35fb82497ec6c5904b0adb7e1c881a76c1c692db 3c62aed89fbe3e15f6f82035e805a0a1add26306 2025-09-30T11:03:23+03:00 Georgi Gerganov metal : dynamic simdgroups for MV kernels (#16340)
|
||
3c62aed89fbe3e15f6f82035e805a0a1add26306 f1eb1cb1eba042b2d583234e80f65e2e225d8995 2025-09-30T09:36:33+02:00 Adrien Gallouët common : simplify etag tracking by removing json (#16342)
|
||
f1eb1cb1eba042b2d583234e80f65e2e225d8995 de41f2b7bffab7582944b213ce12b605f8cf6e0f 2025-09-30T09:07:20+02:00 Charles Xu kleidiai : fix work size and threads sync for fp16 (#16246)
|
||
de41f2b7bffab7582944b213ce12b605f8cf6e0f a74a0d69f34f52fa10d4f0a7ce749fb3490d0774 2025-09-29T22:30:16-07:00 lhez codeowners: add codeowners for opencl backend (#16344)
|
||
a74a0d69f34f52fa10d4f0a7ce749fb3490d0774 5f7e166cbf7b9ca928c7fad990098ef32358ac75 2025-09-29T19:26:34-05:00 Jeff Bolz tests: override test_set_rows::max_nmse_err to allow for occasional rounding differences (#16295)
|
||
5f7e166cbf7b9ca928c7fad990098ef32358ac75 d72f5f7ba260b546190338b0b76f2f152581424f 2025-09-29T18:49:47+02:00 Pascal Fix thinking blocks with quotes + add handling `[THINK]...[/THINK]` blocks (#16326)
|
||
d72f5f7ba260b546190338b0b76f2f152581424f b77e6c18e1a6fac5705ed95f03af5436d67484c1 2025-09-29T17:51:48+03:00 Georgi Gerganov ci : add AMD runners and workflows (#16249)
|
||
b77e6c18e1a6fac5705ed95f03af5436d67484c1 2ddd3f2356c313385fafc19a9f0ce678c8fe03ee 2025-09-29T22:50:44+08:00 alex-spacemit ggml: riscv: add riscv spacemit backend (#15288)
|
||
2ddd3f2356c313385fafc19a9f0ce678c8fe03ee 4d3d455d3c8719eb8f206de328d1b9ba191efd7c 2025-09-29T16:50:52+03:00 Georgi Gerganov sync : ggml
|
||
4d3d455d3c8719eb8f206de328d1b9ba191efd7c c9b1c06467aca8d30fafcab51292bcb285df5b0d 2025-09-29T16:49:11+03:00 Georgi Gerganov sync : whisper.cpp (ggml/1359)
|
||
c9b1c06467aca8d30fafcab51292bcb285df5b0d b6ae75afb49b23db3dfbc2b01e8aabfb047e6880 2025-09-26T17:34:42+02:00 Daniel Bevenius ggml : remove -dev suffix from release version (ggml/1355)
|
||
b6ae75afb49b23db3dfbc2b01e8aabfb047e6880 b6dff20e2fe352093039e2359370c6bb2b505e0b 2025-09-25T14:39:05+02:00 Daniel Bevenius ggml : bump version to 0.9.3 (ggml/1353)
|
||
b6dff20e2fe352093039e2359370c6bb2b505e0b 2db78c75e4ef7497313fda6dbbb3fcaf9ca0752d 2025-09-20T16:44:23+03:00 Georgi Gerganov ggml : prepare for development of 0.9.2-dev
|
||
2db78c75e4ef7497313fda6dbbb3fcaf9ca0752d 02463ab27b1379ff5e3d936ce8b3bfd356872ea6 2025-09-20T16:44:23+03:00 Georgi Gerganov ggml : bump version to 0.9.1
|
||
02463ab27b1379ff5e3d936ce8b3bfd356872ea6 adc76347d73b3d915e946efa5de8d0ad9f3904c2 2025-09-29T13:17:09+02:00 Rafal Lewczuk ggml-backend : add root cause in error message if loading backend library fails (#16172)
|
||
adc76347d73b3d915e946efa5de8d0ad9f3904c2 3a2bdcda0b31e51db954551e0cd49424133a84f3 2025-09-29T11:09:00+02:00 Sigbjørn Skjæret ggml : check cuda and metal argsort limits and add test (#16323)
|
||
3a2bdcda0b31e51db954551e0cd49424133a84f3 66bb7985c3fcefda7a74aae9f8321f70eb1646c4 2025-09-29T10:37:20+02:00 Aleksander Grygier Improve Mobile UI for dialogs and action dropdowns (#16222)
|
||
66bb7985c3fcefda7a74aae9f8321f70eb1646c4 2f61c0f5bf8a620ca4c3872408803ab38cfb9613 2025-09-29T09:08:41+02:00 Pascal fix: preserved zero values in chat settings inputs and textareas by switching to nullish coalescing for field values and default placeholders (#16312)
|
||
2f61c0f5bf8a620ca4c3872408803ab38cfb9613 3ffd0fae473c954bb3e67526b31262048fb508d4 2025-09-29T12:33:12+05:30 Vinkal llama-cli: prevent spurious assistant token (#16202)
|
||
3ffd0fae473c954bb3e67526b31262048fb508d4 a4a0aa5ea2a88e4858996199fefd5439f17b481c 2025-09-29T01:30:45-05:00 ddh0 perplexity : show more kl-divergence data (#16321)
|
||
a4a0aa5ea2a88e4858996199fefd5439f17b481c 92cd103f627015a95f2107f1c27dc218c7b8ec64 2025-09-29T08:41:28+03:00 Georgi Gerganov ggml : fix dependencies for ggml_set_rows (#16318)
|
||
92cd103f627015a95f2107f1c27dc218c7b8ec64 b887d2f3413ac231e3cb5925260c39902af4a70c 2025-09-28T23:50:37-05:00 Jeff Bolz vulkan: Fix validation failure in quantized flash attention (#16292)
|
||
b887d2f3413ac231e3cb5925260c39902af4a70c bd0af02fc96c2057726f33c0f0daf7bb8f3e462a 2025-09-28T23:15:03+02:00 Sigbjørn Skjæret ggml : fix GGML_F32_VEC_FMA argument order in ggml_vec_mad1_f32 (#16307)
|
||
bd0af02fc96c2057726f33c0f0daf7bb8f3e462a d9e0e7c8194dfd7d23bf3a86608c9ece68d77c93 2025-09-28T14:13:50-04:00 crat0z common : fix reasoning before forced tool call via tool_choice = required (#16264)
|
||
d9e0e7c8194dfd7d23bf3a86608c9ece68d77c93 0124ac989f7e7bf08803788f66dbe4106bdcdd58 2025-09-28T22:38:15+08:00 R0CKSTAR ci : fix musa docker build (#16306)
|
||
0124ac989f7e7bf08803788f66dbe4106bdcdd58 2811c65286ae954bec87049f75b86dc022006dcc 2025-09-28T19:25:58+08:00 Aaron Teo devops: switch to using ubuntu-22.04-s390x image (#16302)
|
||
2811c65286ae954bec87049f75b86dc022006dcc d8359f5fde480da030bf75c7711573c7c4d993ba 2025-09-28T12:04:46+01:00 Imad Saddik Fixed a few typos in the README of the LLaMA.cpp HTTP Server [no ci] (#16297)
|
||
d8359f5fde480da030bf75c7711573c7c4d993ba 6a2c6145a0b91b40eb3c3dba7b20ccc4b270490f 2025-09-28T01:38:37-05:00 Jeff Bolz vulkan: 64-bit im2col (#16135)
|
||
6a2c6145a0b91b40eb3c3dba7b20ccc4b270490f 3b53634fe35771e2e318227aa81585726bae7234 2025-09-28T09:34:44+03:00 Georgi Gerganov metal : extend mat-mat multiplication support (#16225)
|
||
3b53634fe35771e2e318227aa81585726bae7234 1384abf8b8d5894d32fada453ccf4d196ffba7de 2025-09-28T09:34:05+03:00 Georgi Gerganov metal : fuse non-sequential nodes (#16102)
|
||
1384abf8b8d5894d32fada453ccf4d196ffba7de e6d65fb02d553bd79cad94e517cdca18b687788d 2025-09-27T20:36:34-05:00 Jeff Bolz vulkan: handle mat_mul with A matrix > 4GB (#16176)
|
||
e6d65fb02d553bd79cad94e517cdca18b687788d 8656f5de688cddcaea1d6174535eb60ee23ef6a0 2025-09-27T16:43:39-04:00 Jeff Bolz vulkan: support arbitrary KV dimension in flash attention (#16160)
|
||
8656f5de688cddcaea1d6174535eb60ee23ef6a0 4807e8f96a61b2adccebd5e57444c94d18de7264 2025-09-27T22:41:03+02:00 Acly vulkan : make the vulkan.hpp dynamic dispatcher instance private (#16224)
|
||
4807e8f96a61b2adccebd5e57444c94d18de7264 c0bfc57af421f8fd63c946c13b7666aed82560e2 2025-09-27T19:56:40+02:00 Aleksander Grygier Show message actions by default (#16289)
|
||
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)
|
||
75a3a6c2cd0002ba40e2dcc92007bc9fdbc69f1a 0499b29c6f64c705faaf5860dc4600fca23671f4 2025-09-27T18:45:07+02:00 Johannes Gäßler CUDA: refactor and deduplicate vector FA kernels (#16208)
|
||
0499b29c6f64c705faaf5860dc4600fca23671f4 234e2ff8ed09716fb553437596779399bee31b11 2025-09-27T19:26:46+03:00 Dmytro Minochkin vulkan: throw system error instead of SIGABRT during init on older devices (#16156)
|
||
234e2ff8ed09716fb553437596779399bee31b11 3f81b4e91c1d5f098148af117e3f13cf4b077f52 2025-09-27T18:17:08+02:00 Adrien Gallouët server : remove old LLAMA_SERVER_SSL (#16290)
|
||
3f81b4e91c1d5f098148af117e3f13cf4b077f52 ace6a54565444b6377bee8e7ac693238e7766279 2025-09-27T06:36:11-04:00 Jeff Bolz vulkan: support GET_ROWS for k-quants (#16235)
|
||
ace6a54565444b6377bee8e7ac693238e7766279 72b24d96c6888c609d562779a23787304ae4609c 2025-09-27T11:12:46+02:00 Adrien Gallouët build : add LLAMA_OPENSSL option (#16287)
|
||
72b24d96c6888c609d562779a23787304ae4609c 624207e676ab5eb3ce7af631902bb45fb73a8359 2025-09-27T02:58:29+05:30 Vinkal model : make minicpm embedding_scale, residual_scale and logit_scale optional with legacy defaults (#16273)
|
||
624207e676ab5eb3ce7af631902bb45fb73a8359 807e8c6d310952f2f5656afc63dab9d7083dcb5c 2025-09-27T02:03:33+08:00 Aaron Teo devops: add s390x & ppc64le CI (#15925)
|
||
807e8c6d310952f2f5656afc63dab9d7083dcb5c 1a189278944d030211f336e103e96b65f976c361 2025-09-26T19:25:29+02:00 Aleksander Grygier Enhance text file detection logic for file attachments (#16199)
|
||
1a189278944d030211f336e103e96b65f976c361 e0539eb6aed346d4b25a6ea019044e88771e7690 2025-09-26T18:35:42+02:00 Aleksander Grygier Allow viewing conversations even when llama server is down (#16255)
|
||
e0539eb6aed346d4b25a6ea019044e88771e7690 5d0a40f390732cbf85d8a3b7b0fc3cbebffe780a 2025-09-26T11:36:48-04:00 Isaac McFadyen webui: switch to hash-based routing (alternative of #16079) (#16157)
|
||
5d0a40f390732cbf85d8a3b7b0fc3cbebffe780a d12a9836597b1ae4440d64d5a6ed727d27ac0702 2025-09-26T15:59:07+02:00 Aleksander Grygier Always show message actions for mobile UI + improvements for user message sizing (#16076)
|
||
d12a9836597b1ae4440d64d5a6ed727d27ac0702 cc1cfa277b3aae1f6cc9180472072336597a78d4 2025-09-26T16:09:34+03:00 Radoslav Gerganov codeowners : add rgerganov as owner of RPC [no ci] (#16279)
|
||
cc1cfa277b3aae1f6cc9180472072336597a78d4 54dbc37053f7d75ea5b0631d5ee88f19391e4314 2025-09-26T15:00:44+02:00 Aleksei Nikiforov mtmd : fix uninitialized variable in bicubic_resize (#16275)
|
||
54dbc37053f7d75ea5b0631d5ee88f19391e4314 b995a10760cb93d23d617d76ecb82a5f95b5e0d3 2025-09-26T14:14:28+03:00 Georgi Gerganov metal : report OOM errors (#16274)
|
||
b995a10760cb93d23d617d76ecb82a5f95b5e0d3 4710dd31bbcef79d04f85a3a6a8c7d9439c5c79a 2025-09-26T13:12:19+02:00 Adrien Gallouët common : use cpp-httplib as a cURL alternative for downloads (#16185)
|
||
4710dd31bbcef79d04f85a3a6a8c7d9439c5c79a 9b26511857ac09ae69ab485168fe2d3ee5fb1d6e 2025-09-26T12:39:35+02:00 Adrien Gallouët build : fix build-ios-device (#16257)
|
||
9b26511857ac09ae69ab485168fe2d3ee5fb1d6e 00217cd41388328c93e9e9644921c4319bb03bcc 2025-09-26T18:27:25+08:00 Aaron Teo ggml-cpu: implement MXFP4 SIMD for s390x (#16193)
|
||
00217cd41388328c93e9e9644921c4319bb03bcc 3b337b01a1a85c2f5b49376e8c0bbdb3f521528e 2025-09-26T13:19:23+03:00 Radoslav Gerganov ci : create git tags for released docker images (#16008)
|
||
3b337b01a1a85c2f5b49376e8c0bbdb3f521528e a86a580a6692edd1da076d5422f944c344b4fe5e 2025-09-26T07:53:36+02:00 Daniel Bevenius codeowners : add danbev as owner of build-xcframework.sh [no ci] (#16268)
|
||
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)
|
||
835b2b915c52bcabcd688d025eacff9a07b65f52 b05a9d650f4da1e70e7e2cbe8fe44e61b39656db 2025-09-25T19:50:28+02:00 Sigbjørn Skjæret model : add GroveMoE support (#15510)
|
||
b05a9d650f4da1e70e7e2cbe8fe44e61b39656db 27052978e487957b3d39e3bd8fc8ee4aa304bff9 2025-09-25T23:38:10+08:00 Aaron Teo vendors: update miniaudio version (#16212)
|
||
27052978e487957b3d39e3bd8fc8ee4aa304bff9 077c94d0caf87fbd3cf3288dbb5c0fd9670294cf 2025-09-25T18:20:34+03:00 rtaluyev readme : update bindings (#16144)
|
||
077c94d0caf87fbd3cf3288dbb5c0fd9670294cf aa3ee0eb0b80efca126cedf9bcb4fb5864b46ce3 2025-09-25T22:35:05+08:00 Aman Gupta CUDA: add a fused top-K MoE kernel (#16130)
|
||
aa3ee0eb0b80efca126cedf9bcb4fb5864b46ce3 d0991da39d3c39b3980c24cdfccb9a4c95a46870 2025-09-25T12:02:36+02:00 Daniel Bevenius model-conversion : add embedding prompt file support (#15871)
|
||
d0991da39d3c39b3980c24cdfccb9a4c95a46870 aa719c2f886c445b78113bf1e5849f1fce4124a7 2025-09-25T11:36:47+02:00 Daniel Bevenius server : add support for external server for tests (#16243)
|
||
aa719c2f886c445b78113bf1e5849f1fce4124a7 4cdd0bb4537f9617e9efcfef6b9454fcefe2ff08 2025-09-25T17:22:55+08:00 junchao-zhao ggml : fix loongarch lsx compilation error (#15864)
|
||
4cdd0bb4537f9617e9efcfef6b9454fcefe2ff08 b5bd037832bcb8ed3086dfe26ce9090bea989af1 2025-09-25T11:12:27+02:00 Johannes Gäßler docs: fix typo [no ci] (#16244)
|
||
b5bd037832bcb8ed3086dfe26ce9090bea989af1 dfcd53f7ecb9bb897a9d752d09c59d10be47237a 2025-09-25T03:53:09-05:00 Douglas Hanley llama : add support for qwen3 reranker (#15824)
|
||
dfcd53f7ecb9bb897a9d752d09c59d10be47237a 4ea00794b8c995b6deaf4bac159c1778dc27419a 2025-09-25T11:30:16+03:00 Georgi Gerganov metal : fuse NORM + MUL + ADD, support non-multiples of 4 (#16220)
|
||
4ea00794b8c995b6deaf4bac159c1778dc27419a 02a6a82ae7c7ddd1819ae26b0cc36675879aecee 2025-09-25T11:29:42+03:00 Georgi Gerganov metal : relax reorder conditions (#16216)
|
||
02a6a82ae7c7ddd1819ae26b0cc36675879aecee c498fc82fe5b83fc8c6e1627286bdc1f93caddbf 2025-09-25T11:29:08+03:00 Georgi Gerganov metal : restore im2col perf (#16219)
|
||
c498fc82fe5b83fc8c6e1627286bdc1f93caddbf e7a5130a20cf45a3358308356c249734ca982143 2025-09-25T10:20:02+03:00 Radoslav Gerganov rpc : use ggml logging facilities
|
||
e7a5130a20cf45a3358308356c249734ca982143 bee378e0988b44bbe93b0768208080951b312363 2025-09-25T13:06:30+08:00 Aaron Teo codeowners: add ownership of zdnn backend [no ci] (#16232)
|
||
bee378e0988b44bbe93b0768208080951b312363 5fb557653b8756ac2b6c6a102b1e44abcadf552f 2025-09-25T05:06:06Z Eve ci: run the x64 and arm ci on the github machines instead (#16183)
|
||
5fb557653b8756ac2b6c6a102b1e44abcadf552f 4ae88d07d026e66b41e85afece74e88af54f4e66 2025-09-25T11:36:30+08:00 Aaron Teo devops: fix s390x docker release failure (#16231)
|
||
4ae88d07d026e66b41e85afece74e88af54f4e66 e789095502b337690c7616db32d7c679a5bd2533 2025-09-25T00:25:04+08:00 Aaron Teo codeowners: add ownership of zdnn backend [no ci] (#16229)
|
||
e789095502b337690c7616db32d7c679a5bd2533 f2a789e33490deb483a2694b066b37e45524bb79 2025-09-24T16:53:48+02:00 Johannes Gäßler llama: print memory breakdown on exit (#15860)
|
||
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)
|
||
63b54c81a620981be020184ab99e63a8e50e47cb 152729f8848b692e1ae53c197ccca92ef10a523b 2025-09-24T16:25:26+08:00 Jie Fu (傅杰) model-conversion : make causal-verify-logits fails with model names containing "." (#16215)
|
||
152729f8848b692e1ae53c197ccca92ef10a523b c0c59c1157b5aeded285a3b25d2463c866f583d3 2025-09-24T08:53:47+02:00 Uilian Ries common : add missing chrono header for common.cpp (#16211)
|
||
c0c59c1157b5aeded285a3b25d2463c866f583d3 7735706b939847df2c6c00b44c36f95a20137c61 2025-09-24T08:53:20+02:00 Sigbjørn Skjæret codeowners : match all requirements files (#16214)
|
||
7735706b939847df2c6c00b44c36f95a20137c61 4d9ea03d170cf504a40bf3a85788af84cccaee80 2025-09-24T14:46:52+08:00 Jie Fu (傅杰) model-conversion : run-org-model.py fails to run on mac m1 (#16213)
|
||
4d9ea03d170cf504a40bf3a85788af84cccaee80 8ba548dae251f8f2e3834ed5226b76b6e4f4a485 2025-09-24T08:10:09+02:00 Daniel Bevenius codeowners : use slash prefix for root files [no ci] (#16210)
|
||
8ba548dae251f8f2e3834ed5226b76b6e4f4a485 f505bd83ca7a43c4585ff3d59135e77eae9c793b 2025-09-24T12:19:23+08:00 Jie Fu (傅杰) model-conversion : fix the make targets in the README.md (#16209)
|
||
f505bd83ca7a43c4585ff3d59135e77eae9c793b 0889589dbe8c67dd518c58a57afbb02dde2dccbe 2025-09-23T20:41:40+03:00 Georgi Gerganov ci : disable AMD workflows + update NVIDIA workflows (#16200)
|
||
0889589dbe8c67dd518c58a57afbb02dde2dccbe 4e29084ba4104c4ea529fd3163bb6e76f64383df 2025-09-23T13:44:25+03:00 Georgi Gerganov ci : enable Vulkan workflow on Mac (#16194)
|
||
4e29084ba4104c4ea529fd3163bb6e76f64383df f6b4af3d04763b1e0130f5b5fce19c4bc6f83f1c 2025-09-23T01:58:12-07:00 Xiangyan Sun ggml-cpu: Respect cpumask settings (#16164)
|
||
f6b4af3d04763b1e0130f5b5fce19c4bc6f83f1c 264f1b51872c125e23fa0ac1da5e2a1170de9a08 2025-09-23T10:25:20+02:00 Sigbjørn Skjæret ggml : fix uninitialized is_on_grid in quantize_row_iq3_xxs_impl (#15928)
|
||
264f1b51872c125e23fa0ac1da5e2a1170de9a08 0bc7cc715472fe28822e57f036f0746592ed2c04 2025-09-23T14:53:05+08:00 Aaron Teo zdnn: refactor codebase + add docs (#16178)
|
||
0bc7cc715472fe28822e57f036f0746592ed2c04 4b9f4cb0f89a88de4bdf97727d0457b0c648804c 2025-09-23T08:13:22+02:00 Daniel Bevenius codeowners : add @danbev to model-conversion example [no ci] (#16190)
|
||
4b9f4cb0f89a88de4bdf97727d0457b0c648804c 85e72271ba1ce78adf34fd8997803c991e617ca6 2025-09-23T13:59:34+08:00 Aaron Teo devops: add s390x containers (#15915)
|
||
85e72271ba1ce78adf34fd8997803c991e617ca6 1d0125bcf1cbd7195ad0faf826a20bc7cec7d3f4 2025-09-23T05:59:03+02:00 Daniel Bevenius ggml-cpu : fix typo in gemm comments [no ci] (#16189)
|
||
1d0125bcf1cbd7195ad0faf826a20bc7cec7d3f4 351f3da39c85f59d581fc184f09283da7f099a3b 2025-09-22T12:40:10-06:00 Gabe Goodhart feat: Add conversion support in GraniteHybrid for non-hybrid (all attn) (#16177)
|
||
351f3da39c85f59d581fc184f09283da7f099a3b 3ecb2f671a2f49d56357f99d135a94e841759178 2025-09-23T01:57:46+08:00 Haiyue Wang clang-tidy : disable warning about performance enum size (#16127)
|
||
3ecb2f671a2f49d56357f99d135a94e841759178 432cf4304c5379bfbd1f3feed65ec205955b2f4b 2025-09-22T19:13:00+02:00 Sigbjørn Skjæret ggml : implement set_rows with i32 index (#16159)
|
||
432cf4304c5379bfbd1f3feed65ec205955b2f4b 37a23c17bdc9c99c9c6ad41168e4ced3724b72cd 2025-09-22T18:20:21+03:00 Georgi Gerganov codeowners : update + cleanup (#16174)
|
||
37a23c17bdc9c99c9c6ad41168e4ced3724b72cd 138c87ce8bd558b2cc134ada7316a3dad8eb67ac 2025-09-22T14:13:51+02:00 Adrien Gallouët common : enable `--offline` mode without curl support (#16137)
|
||
138c87ce8bd558b2cc134ada7316a3dad8eb67ac c6db9a10278f40b4b8d3f6931728f2cce2356daa 2025-09-22T10:53:13+02:00 Quentin Bramas webui : fix handling incomplete chunks (#16107)
|
||
c6db9a10278f40b4b8d3f6931728f2cce2356daa d05affbab7d1364fa7ac06c03cd33f03235ae840 2025-09-22T10:49:58+02:00 GideonSerf embedding : fix typos in README (#16171)
|
||
d05affbab7d1364fa7ac06c03cd33f03235ae840 4f324a556caa5be0be2261604ef4aab0faf36eb8 2025-09-22T16:48:42+08:00 Haiyue Wang common : remove unused local variables (#16140)
|
||
4f324a556caa5be0be2261604ef4aab0faf36eb8 a71ae3ba7a857394103f12e30b387c48242c84f2 2025-09-22T11:12:37+03:00 Georgi Gerganov ggml : extend ggml_can_fuse to work with non-sequential nodes (#16123)
|
||
a71ae3ba7a857394103f12e30b387c48242c84f2 05a2458121f9cdcc98ea6ffeeab6f23023136266 2025-09-22T11:12:09+03:00 Georgi Gerganov ggml : add ggml_op_is_empty (#16122)
|
||
05a2458121f9cdcc98ea6ffeeab6f23023136266 96fdca043be8fa733bbd59868a75517f92633376 2025-09-22T15:10:58+07:00 Xuan-Son Nguyen codeowners : update ownership for @ngxson and @allozuar (#16128)
|
||
96fdca043be8fa733bbd59868a75517f92633376 b2d980fce063fa3591c59ad50b946c8da66c294f 2025-09-22T17:04:01+09:00 Shin-myoung-serp Vulkan: add conv_transpose_2d operation (#16022)
|
||
b2d980fce063fa3591c59ad50b946c8da66c294f 5c6106a696af2b00ce01d14ba525979d5a32b199 2025-09-22T09:59:05+02:00 Sigbjørn Skjæret codeowners : claim responsibility for ci, models, gguf-py and convert (#16124)
|
||
5c6106a696af2b00ce01d14ba525979d5a32b199 ec65fb52f0cc890e5085a6c5995ab08a156265fb 2025-09-22T10:58:02+03:00 Georgi Gerganov contrib : update roles (#16113)
|
||
ec65fb52f0cc890e5085a6c5995ab08a156265fb 1d660d2fae42ea2e1d3569638e722bf7a37b6b19 2025-09-22T10:16:05+03:00 Georgi Gerganov ci : remove vulkaninfo calls (#16169)
|
||
1d660d2fae42ea2e1d3569638e722bf7a37b6b19 a20d810d79adfbf82e0eb703af6f27a0e2d1a539 2025-09-22T09:11:39+03:00 Georgi Gerganov ci : use smaller model (#16168)
|
||
a20d810d79adfbf82e0eb703af6f27a0e2d1a539 4d0a7cbc617e384fc355077a304c883b5c7d4fb6 2025-09-22T00:37:17-05:00 Jeff Bolz vulkan: add RTE variants of exp shader (#16165)
|
||
4d0a7cbc617e384fc355077a304c883b5c7d4fb6 9073a73d82a916cea0809de225ef5175c3a86e91 2025-09-22T08:31:40+03:00 Georgi Gerganov ci : adjust params for less runtime (#16167)
|
||
9073a73d82a916cea0809de225ef5175c3a86e91 51f5a45fbe575dcd54bdd2a339ef8e8424d1c12a 2025-09-22T07:22:43+02:00 Ruben Ortlam vulkan: vec dot matrix multiplication fix (#16151)
|
||
51f5a45fbe575dcd54bdd2a339ef8e8424d1c12a c4510dc9374e17dcb8726902ab5216067a92b3d3 2025-09-21T16:42:10-07:00 lhez opencl: fix concat crash on win arm64 with Adreno (#15944)
|
||
c4510dc9374e17dcb8726902ab5216067a92b3d3 da30ab5f8696cabb2d4620cdc0aa41a298c54fd6 2025-09-21T14:48:44-07:00 lhez opencl: initial `q8_0` mv support (#15732)
|
||
da30ab5f8696cabb2d4620cdc0aa41a298c54fd6 28baac9c9f491c872e2c37762d3bd90446b005e9 2025-09-21T19:00:27+03:00 Georgi Gerganov ci : add label for the RISC-V runner (#16150)
|
||
28baac9c9f491c872e2c37762d3bd90446b005e9 1eeb523c3e0c7ffbd59469f5463dcbdecba3535e 2025-09-21T16:50:45+03:00 Georgi Gerganov ci : migrate ggml ci to self-hosted runners (#16116)
|
||
1eeb523c3e0c7ffbd59469f5463dcbdecba3535e 5bb4a3edec297e74b0f7bd4ed5d0fdd12e28d858 2025-09-21T08:31:55+02:00 Giuseppe Scrivano vulkan: optimize UMA buffer operations and fix driver hangs (#16059)
|
||
5bb4a3edec297e74b0f7bd4ed5d0fdd12e28d858 7f766929ca8e8e01dcceb1c526ee584f7e5e1408 2025-09-21T01:23:37-05:00 Jeff Bolz vulkan: fix validation error about VK_PIPELINE_CREATE_CAPTURE_STATISTICS_BIT_KHR (#16086)
|
||
7f766929ca8e8e01dcceb1c526ee584f7e5e1408 405921dcefdd4e90ed948a4bf179007c2fa92b2d 2025-09-20T12:55:47+03:00 Georgi Gerganov sync : ggml
|
||
405921dcefdd4e90ed948a4bf179007c2fa92b2d fa6383ca7e7ccb8ca3bdfeb37e348ddc4113aa26 2025-09-16T06:16:52+02:00 Daniel Bevenius ggml : introduce semantic versioning (ggml/1336)
|
||
fa6383ca7e7ccb8ca3bdfeb37e348ddc4113aa26 803dac2e48ef3ba26a504eb27c4e77ec2d21f7d0 2025-09-10T17:21:11+02:00 Gregor Jasny CUDA : conditionally add cuda architectures (ggml/1341)
|
||
803dac2e48ef3ba26a504eb27c4e77ec2d21f7d0 459c0c2c1a400f960d7b8e8d94d31a8426f80986 2025-09-20T10:42:56+02:00 Ruben Ortlam vulkan: use vec dot for matrix matrix multiplications (#16056)
|
||
459c0c2c1a400f960d7b8e8d94d31a8426f80986 be79d9fdd95ab8955527c4aaa67b90e8b9516718 2025-09-20T07:56:30+02:00 Benni server: fix SSE and OpenAI compatibility for error messages when streaming (#16109)
|
||
be79d9fdd95ab8955527c4aaa67b90e8b9516718 f432d8d83e7407073634c5e4fd81a3d23a10827f 2025-09-19T15:15:21-07:00 ssweens llama-bench: add --devices and --list-devices support (#16039)
|
||
f432d8d83e7407073634c5e4fd81a3d23a10827f 4067f07fc5aa5722b87db303b561597004696f6c 2025-09-20T00:57:30+09:00 shun095 chat: Fix streaming parser for granite models (#15682)
|
||
4067f07fc5aa5722b87db303b561597004696f6c 4b8560ab56fdd9819358b47c338bbc8ec357c57e 2025-09-19T09:52:27+02:00 Aleksander Grygier feat: Improve mobile UI for Settings Dialog (#16084)
|
||
4b8560ab56fdd9819358b47c338bbc8ec357c57e 0dd58b6877f3dc106593d6bc68d98305f553c566 2025-09-19T13:02:51+07:00 Xuan-Son Nguyen chat : fix build on arm64 (#16101)
|
||
0dd58b6877f3dc106593d6bc68d98305f553c566 69ffd891631befa9e6b485fd646a16dab4f2c007 2025-09-19T11:31:56+07:00 Xuan-Son Nguyen ggml : refactor forward_dup for cpu backend (#16062)
|
||
69ffd891631befa9e6b485fd646a16dab4f2c007 246c0d9c795fb25da8268625d597580155a3672f 2025-09-18T23:07:26+02:00 Adrien Gallouët ggml-amx : fix ggml_amx_init() on generic Linux (#16049)
|
||
246c0d9c795fb25da8268625d597580155a3672f 3edd87cd055a45d885fa914d879d36d33ecfc3e1 2025-09-18T23:07:18+02:00 Adrien Gallouët cmake : fix static linking for OpenMP on Unix-like systems (#16031)
|
||
3edd87cd055a45d885fa914d879d36d33ecfc3e1 c0b45097c33e2667a94444f08cc9e36bec0a5e2e 2025-09-18T12:03:34-07:00 Shawn Gu opencl: optimize mxfp4 kernels (#16037)
|
||
c0b45097c33e2667a94444f08cc9e36bec0a5e2e 38dbdf4c057515ccea9bec0ca2518f86d5e4d28e 2025-09-18T13:46:17-05:00 Jeff Bolz rename optimize_graph to graph_optimize (#16082)
|
||
38dbdf4c057515ccea9bec0ca2518f86d5e4d28e 368560a1e3b9a3bc83af741b0b2bc9e46fb420d2 2025-09-18T11:26:03-07:00 Bowen Han CUDA: Optimize PAD_REFLECT_1D (#15957)
|
||
368560a1e3b9a3bc83af741b0b2bc9e46fb420d2 4ca088b036313c2d8e682f4cfeb7c29edd85d0b9 2025-09-18T19:28:32+02:00 Johannes Gäßler CUDA: fix compilation on CC 6.0 (#16091)
|
||
4ca088b036313c2d8e682f4cfeb7c29edd85d0b9 703f9e32c4eb3166f8d63007c26e31a1466c21af 2025-09-18T16:22:50+01:00 Eric Curtin Add resumable downloads for llama-server model loading (#15963)
|
||
703f9e32c4eb3166f8d63007c26e31a1466c21af ad6bd9083bc346afd4ecd2fbb83e0306c0141c99 2025-09-18T16:28:41+03:00 Georgi Gerganov metal : use function constants for mul_mv_ext kernels (#16074)
|
||
ad6bd9083bc346afd4ecd2fbb83e0306c0141c99 2b6b55a59f590a1e1eb2dcd09d5b8b6feb9cc748 2025-09-18T13:28:22+02:00 Sigbjørn Skjæret cuda : add missing F32<->I32 entries in ggml_cuda_cpy_fn (#16060)
|
||
2b6b55a59f590a1e1eb2dcd09d5b8b6feb9cc748 e58174cecbc45bf79bf653cd2c984395940c6ef4 2025-09-18T13:36:57+03:00 Radoslav Gerganov server : include usage statistics only when user request them (#16052)
|
||
e58174cecbc45bf79bf653cd2c984395940c6ef4 b213fce89bee8cb56b587b91e15a4278f8ed0180 2025-09-18T12:47:56+03:00 Georgi Gerganov llama : bump max seq limit from 64 to 256 (#15916)
|
||
b213fce89bee8cb56b587b91e15a4278f8ed0180 e00f3fd8fff2cf5a8c8c9f475034bd089c8bcce4 2025-09-18T12:33:45+03:00 Georgi Gerganov metal : improve F32, F16 and BF16 mat-vec multiplication (#16057)
|
||
e00f3fd8fff2cf5a8c8c9f475034bd089c8bcce4 f2f28380ea68939c0481df3e4216b698824a59df 2025-09-18T15:06:48+08:00 Jhen-Jie Hong metal : avoid call free for non-owned buffer (#16067)
|
||
f2f28380ea68939c0481df3e4216b698824a59df 62c3b645c56ad5288aa401901f043b050edac5a2 2025-09-18T10:03:24+03:00 Georgi Gerganov metal : handle nil cv during pipeline creation (#16065)
|
||
62c3b645c56ad5288aa401901f043b050edac5a2 d304f459d8619399eb232b1e3689379306f439d3 2025-09-18T09:26:33+08:00 Chenguang Li CANN: Remove print (#16044)
|
||
d304f459d8619399eb232b1e3689379306f439d3 0320ac5264279d74f8ee91bafa6c90e9ab9bbb91 2025-09-17T13:09:40-07:00 Reese Levine GGML WebGPU: Support for ADD, MUL, RMS_NORM, GET_ROWS operators (#16018)
|
||
0320ac5264279d74f8ee91bafa6c90e9ab9bbb91 a7a98e0fffed794396b3fbad4dcdbbc184963645 2025-09-17T20:38:12+03:00 Georgi Gerganov metal : refactor + optimize v2 (#15995)
|
||
a7a98e0fffed794396b3fbad4dcdbbc184963645 8f8f2274ee3601fecf6e2d57b52f701c81bede21 2025-09-17T19:29:13+02:00 Aleksander Grygier SvelteKit-based WebUI (#14839)
|
||
8f8f2274ee3601fecf6e2d57b52f701c81bede21 c959b676be29e93f8dbc3bd6056ceba812a9eb72 2025-09-18T00:18:21+07:00 Xuan-Son Nguyen convert : add Llama4ForCausalLM (#16042)
|
||
c959b676be29e93f8dbc3bd6056ceba812a9eb72 cd08fc3ecc0264b4414b68af3874a6c689ed60c1 2025-09-17T15:32:42+02:00 Johannes Gäßler CUDA: fix FA occupancy, optimize tile kernel (#15982)
|
||
cd08fc3ecc0264b4414b68af3874a6c689ed60c1 cb5bb6cc05119c24e7711ca2956cd0e6d409d396 2025-09-17T01:08:02-07:00 David Ribeiro Alves common : Fix corrupted memory error on json grammar initialization (#16038)
|
||
cb5bb6cc05119c24e7711ca2956cd0e6d409d396 a91d035b901e8a9edf810f63d130ee49adf27be2 2025-09-17T07:35:37Z Eve vulkan: automatically remove unsupported devices (#15976)
|
||
a91d035b901e8a9edf810f63d130ee49adf27be2 745cbcf2fe1eb88f8db615ac622f0b944d924ad6 2025-09-17T09:34:09+02:00 Daniel Bevenius ci : revert back to macos-13 for macOS-latest-cmake-x64 (#16040)
|
||
745cbcf2fe1eb88f8db615ac622f0b944d924ad6 1cbd80f8cf80a817715b1ccc5680fe2a3c5172c8 2025-09-17T15:30:55+08:00 Jie Fu (傅杰) llama-quant : fix the verification of attention layers for encoder-decoder models (#16023)
|
||
1cbd80f8cf80a817715b1ccc5680fe2a3c5172c8 85286f354813056f6c835046c0acfa3bf6ba9432 2025-09-17T15:29:00+08:00 Jie Fu (傅杰) examples : support encoder-decoder models in the simple example (#16002)
|
||
85286f354813056f6c835046c0acfa3bf6ba9432 d5fabe3682de515fd09d6c981f7a0d1b75614455 2025-09-17T00:01:58-07:00 Shane A model : add OLMo3 support (#16015)
|
||
d5fabe3682de515fd09d6c981f7a0d1b75614455 8ff206097c2bf3ca1c7aa95f9d6db779fc7bdd68 2025-09-17T14:33:08+08:00 Chenguang Li CANN: Optimize ggml_cann_set_device (#15935)
|
||
8ff206097c2bf3ca1c7aa95f9d6db779fc7bdd68 77475530b8bbea3bf578632507e1284cdfe2c8c0 2025-09-16T16:17:08+02:00 jacekpoplawski llama-bench: add --n-cpu-moe support (#15952)
|
||
77475530b8bbea3bf578632507e1284cdfe2c8c0 3913f8730ec6d6245480affc30ae3049107956f4 2025-09-16T15:27:52+02:00 Daniel Bevenius ci : use macos-latest for arm64 webgpu build (#16029)
|
||
3913f8730ec6d6245480affc30ae3049107956f4 76888d202ed2b835ae19ea9f9db6baf39e419297 2025-09-16T15:25:57+02:00 Daniel Bevenius ggml : fix padding in timestep embedding kernels (#15932)
|
||
76888d202ed2b835ae19ea9f9db6baf39e419297 f1fbffb5c0b34b2a68febb7da3fd0f8333f1ed4c 2025-09-16T13:41:38+02:00 Daniel Bevenius ci : upload xcframework artifact from ios-xcode-build job (#16010)
|
||
f1fbffb5c0b34b2a68febb7da3fd0f8333f1ed4c 51abc96bdc52ba8cd6ad78dcf12ed9a041d7b442 2025-09-15T23:59:19-07:00 Bowen Han fix: apply clang-format to CUDA macros (#16017)
|
||
51abc96bdc52ba8cd6ad78dcf12ed9a041d7b442 07808ebb07e2b1aa19032705e332679ddf967614 2025-09-16T05:57:16+02:00 Daniel Bevenius ci : update macos-latest* jobs to use macos-latest (#15938)
|
||
07808ebb07e2b1aa19032705e332679ddf967614 6d758839ff741d4966ca92b7f801b7a8b5b96364 2025-09-15T22:54:44-04:00 Yuri Khrustalev cmake : Do not install tools on iOS targets (#15903)
|
||
6d758839ff741d4966ca92b7f801b7a8b5b96364 3d4053f77f0f78ee2b791088c02af653ebee42dd 2025-09-16T10:38:28+08:00 Aman Gupta Add LLaDA-7b-MoE diffusion model (#16003)
|
||
3d4053f77f0f78ee2b791088c02af653ebee42dd dc381aa9a6dc45f00673471d34b8bddd30e77570 2025-09-15T16:28:31-06:00 Jake Karnes CUDA: fix im2col_3d to respect non-contiguous inputs (views) (#15956)
|
||
dc381aa9a6dc45f00673471d34b8bddd30e77570 10d197409bd9537ff302ad09966fe406882fef9d 2025-09-15T14:38:52-07:00 Diego Devesa docker : enable rocWMMA in ROCm images, add gfx1151 (#15997)
|
||
10d197409bd9537ff302ad09966fe406882fef9d b907255f4bd169b0dc7dca9553b4c54af5170865 2025-09-15T14:38:42-07:00 Diego Devesa releases : switch to rocWMMA develop branch, add gfx1151 (#15992)
|
||
b907255f4bd169b0dc7dca9553b4c54af5170865 28c39da7c645185ade5436767929d7ec33006033 2025-09-15T19:51:35+03:00 yael-works SYCL: Add COUNT_EQUAL operator support (#15991)
|
||
28c39da7c645185ade5436767929d7ec33006033 106220562aca42b6738b8f51acfce0db1b8a2fb6 2025-09-15T13:08:30+03:00 Nikolay Popov llama-run: Fix model download on Windows (#15988)
|
||
106220562aca42b6738b8f51acfce0db1b8a2fb6 a68f31edd71cc39141113f05f7133a3e9ece8c61 2025-09-15T17:35:11+08:00 Aman Gupta CUDA: some micro-optimizations in mmf.cuh for mul_mat_id (#15926)
|
||
a68f31edd71cc39141113f05f7133a3e9ece8c61 b8e09f08b9a91c0401bc67d17a17c90756420346 2025-09-15T02:54:57-05:00 ddh0 fix KLD percentile output (#15999)
|
||
b8e09f08b9a91c0401bc67d17a17c90756420346 6c019cb04e86e2dacfe62ce7666c64e9717dde1f 2025-09-14T23:00:59+02:00 Sigbjørn Skjæret model : add grok-2 support (#15539)
|
||
6c019cb04e86e2dacfe62ce7666c64e9717dde1f 9dcd200d57bc6f05a59a6a8df361d5d183af4124 2025-09-14T21:17:04+02:00 Sigbjørn Skjæret server : only attempt to enable thinking if using jinja (#15967)
|
||
9dcd200d57bc6f05a59a6a8df361d5d183af4124 0fa154e3502e940df914f03b41475a2b80b985b0 2025-09-14T22:02:32+03:00 Georgi Gerganov metal : remove memory pools (#15966)
|
||
0fa154e3502e940df914f03b41475a2b80b985b0 261e6a20ffdb79c4875e674b4f6b514bc73cff8f 2025-09-15T04:43:54+10:00 Adam rocm.Dockerfile: added gfx1200,gfx1201 architectures to support AMD Radeon RX 9000 series (#15994)
|
||
261e6a20ffdb79c4875e674b4f6b514bc73cff8f a0e13dcbe5bae7025660349ef3e4ead060e507f2 2025-09-14T16:56:28+02:00 Ruben Ortlam Vulkan: Clean up mul_mm shader (#15987)
|
||
a0e13dcbe5bae7025660349ef3e4ead060e507f2 a14bd350141fb42b8bf2dd2342cebc27bfdce399 2025-09-14T22:20:35+08:00 lcy build: fix the build failures of Windows HIP release job (#15984)
|
||
a14bd350141fb42b8bf2dd2342cebc27bfdce399 918b26f197f55d5d562446dfc876d0e637929d07 2025-09-14T15:33:22+03:00 Georgi Gerganov metal : fix kernel requirements (#15983)
|
||
918b26f197f55d5d562446dfc876d0e637929d07 9ecb88434644c865232bb665d2f6f05049fc6456 2025-09-14T12:28:18+03:00 Radoslav Gerganov rpc : fix regression when --device is used (#15981)
|
||
9ecb88434644c865232bb665d2f6f05049fc6456 d1c6f11f47bae063a68a6be8e4830c060a11b7bd 2025-09-14T02:21:59-07:00 Diego Devesa releases : update ROCM, add gfx1200, gfx1201, gfx1151 (#15972)
|
||
d1c6f11f47bae063a68a6be8e4830c060a11b7bd 6380d6a3e709cb02a8695afdd96b40e674477332 2025-09-14T12:10:07+03:00 Radoslav Gerganov doc : update documentation for --tensor-split (#15980)
|
||
6380d6a3e709cb02a8695afdd96b40e674477332 aa0c461efe3603639af1a1defed2438d9c16ca0f 2025-09-14T13:37:03+08:00 Aaron Teo ggml-zdnn: rm user mapped buffers (#15965)
|
||
aa0c461efe3603639af1a1defed2438d9c16ca0f b9c9c9f789cd57fb0b28b25223305613cd90fa10 2025-09-13T16:29:43+01:00 Jeff Bolz vulkan: fix failing dequant shaders (#15862)
|
||
b9c9c9f789cd57fb0b28b25223305613cd90fa10 50f4281a6f5c3a5d68bdeb12f904fa01e0e2ba91 2025-09-13T16:23:30+01:00 Jeff Bolz vulkan: initialize vulkan-hpp to allow using extension function pointers (#15705)
|
||
50f4281a6f5c3a5d68bdeb12f904fa01e0e2ba91 55758b00cae1451a0d789c50a5eb8c9bee42b9a0 2025-09-13T07:49:49-07:00 Diego Devesa llama : allow using iGPUs with --device (#15951)
|
||
55758b00cae1451a0d789c50a5eb8c9bee42b9a0 f161463a54d9f93d41246286aa4a9569a91d804d 2025-09-13T16:24:22+03:00 Georgi Gerganov metal : refactor kernel loading (#15964)
|
||
f161463a54d9f93d41246286aa4a9569a91d804d 84d7b2fca11d1be118ce776f6d72a486c4883b74 2025-09-13T13:54:28+03:00 Georgi Gerganov metal : allow ops to run concurrently (#15929)
|
||
84d7b2fca11d1be118ce776f6d72a486c4883b74 40be51152d4dc2d47444a4ed378285139859895b 2025-09-13T12:45:04+03:00 Georgi Gerganov metal : fix memory leaks (#15962)
|
||
40be51152d4dc2d47444a4ed378285139859895b 4bf5549269d99bac936a65892da2312cc71b2421 2025-09-13T02:39:52+08:00 Aaron Teo ggml-zdnn: fix #15414, activate FP16 and BF16 acceleration and incorrect zTensor free (#15839)
|
||
4bf5549269d99bac936a65892da2312cc71b2421 f4e664f838cc65d89fa845c48c372e43852112e4 2025-09-12T16:31:50+01:00 Eric Curtin Add docker protocol support for llama-server model loading (#15790)
|
||
f4e664f838cc65d89fa845c48c372e43852112e4 f088b6a84f6aed0abb619dbb9d375e726fc888a6 2025-09-12T23:16:32+08:00 Haiyue Wang context : remove redundant explicit casting to the same type (#15948)
|
||
f088b6a84f6aed0abb619dbb9d375e726fc888a6 304ac5693d1e2124f83e0584bc5eea6311d3d3b4 2025-09-12T17:02:55+03:00 Georgi Gerganov server : adjust prompt similarity thold + add logs (#15913)
|
||
304ac5693d1e2124f83e0584bc5eea6311d3d3b4 6c88ad8fa741a2182a2dcf8c5353ae4b5c66e656 2025-09-12T13:24:21+02:00 Ruben Ortlam Vulkan iGPU device selection overhaul and PCI ID API support (#15947)
|
||
6c88ad8fa741a2182a2dcf8c5353ae4b5c66e656 704d90c987cdf00751567b2088c4e54742aa2d3f 2025-09-12T09:06:20+02:00 Mathieu Baudier vulkan: Make device memory check more portable (#15939)
|
||
704d90c987cdf00751567b2088c4e54742aa2d3f 360d6533db39e11577afe9b0aece20c6b5ddaf1f 2025-09-12T09:15:12+08:00 Neo Zhang Jianyu Revert "sycl: add usage of enqueue_functions extension (#14244)" (#15910)
|
||
360d6533db39e11577afe9b0aece20c6b5ddaf1f 0e6ff0046f4a2983b2c77950aa75960fe4b4f0e2 2025-09-11T13:47:38-07:00 Diego Devesa ggml-backend : add GGML_BACKEND_DEVICE_TYPE_IGPU device type (#15797)
|
||
0e6ff0046f4a2983b2c77950aa75960fe4b4f0e2 df082f56309073ecf885eceaa21b86e8a487e61b 2025-09-11T21:19:58+02:00 Johannes Gäßler CUDA: larger SRAM reads for tile FA, AMD FP16 dot (#15927)
|
||
df082f56309073ecf885eceaa21b86e8a487e61b 24a6734daf6932ff29ba8c1ff0245c51d76f783e 2025-09-11T12:12:34-05:00 ddh0 nitpick : correct MB to MiB (#15934)
|
||
24a6734daf6932ff29ba8c1ff0245c51d76f783e 2b3efea9a4d91216850856fbb77075db26f6a6eb 2025-09-11T15:39:12+02:00 Daniel Bevenius ggml-cpu : add check for ARM MATMUL_INT8/i8mm support (#15922)
|
||
2b3efea9a4d91216850856fbb77075db26f6a6eb c0389dba43d50695f9d3f57dd1f1a14cbefc100c 2025-09-11T12:45:40+02:00 Charles Xu kleidiai: fix GGML_ASSERT(*cur_backend_id != -1) failed (#15614)
|
||
c0389dba43d50695f9d3f57dd1f1a14cbefc100c 00681dfc16ba4cebb9c7fbd2cf2656e06a0692a4 2025-09-11T15:59:37+08:00 hipudding CANN: Disable acl_graph for prefill stage (#15933)
|
||
00681dfc16ba4cebb9c7fbd2cf2656e06a0692a4 4f658855fa8f2e42b7ed9a5b298fa39a2e39b096 2025-09-10T22:04:03+02:00 Oliver Simons CUDA: Add `fastdiv` to `k_bin_bcast*`, giving 1-3% E2E performance (#15872)
|
||
4f658855fa8f2e42b7ed9a5b298fa39a2e39b096 6ab397e12ba8e9f776341cdae68f7ffb2f8d2cde 2025-09-11T02:51:51+08:00 Jie Fu (傅杰) llama : support T5 models with unequal number of encoder-decoder layers (#15909)
|
||
6ab397e12ba8e9f776341cdae68f7ffb2f8d2cde 9de447d94e1ae9d1a36e5a2e5bf47483352c0d9c 2025-09-10T19:08:59+02:00 Sigbjørn Skjæret graph : support non-contiguous Q in build_attn_mha (#15908)
|
||
9de447d94e1ae9d1a36e5a2e5bf47483352c0d9c 0f0a3c2851134d49955f3c85afbb0b1bb47c3e07 2025-09-10T17:31:40+02:00 Daniel Bevenius ggml-cpu : fix padding in ggml_timestep_embedding (#15917)
|
||
0f0a3c2851134d49955f3c85afbb0b1bb47c3e07 33daece86b65607451d0d4378d2d04ba6a20ad55 2025-09-10T17:52:35+03:00 Georgi Gerganov metal : make the backend async (#15906)
|
||
33daece86b65607451d0d4378d2d04ba6a20ad55 e7b6d83b524bbc24a1343d862de6dba8e8eddbd6 2025-09-10T15:39:57+02:00 Daniel Bevenius ci : add caching for ROCm installation in release workflow (#15924)
|
||
e7b6d83b524bbc24a1343d862de6dba8e8eddbd6 2cfef4d117d67ab1dec002915b48a15d11ee1973 2025-09-10T14:17:09+02:00 Daniel Bevenius tests : filter out no-ops from coverage report (#15900)
|
||
2cfef4d117d67ab1dec002915b48a15d11ee1973 09e72a037c77b6823fde9e1e4cf28e815b9c41ab 2025-09-10T12:51:28+01:00 j-k media : add transparent icon svg and png [no ci] (#15891)
|
||
09e72a037c77b6823fde9e1e4cf28e815b9c41ab 10d8b2b6b0ac2cae252a80b4daea5da55ab63c2f 2025-09-10T07:28:47-04:00 Jesse gitignore : Ignore vim swap files in tests (#15901)
|
||
10d8b2b6b0ac2cae252a80b4daea5da55ab63c2f 28b5f190ef1dbea5edf82dbc8b4407b721fadd13 2025-09-10T18:42:00+08:00 Chenguang Li CANN: Add ROPE sin/cos cache for reuse (#15912)
|
||
28b5f190ef1dbea5edf82dbc8b4407b721fadd13 86587da03bd78df8f4e7d8b111a0c1d2494d6ed0 2025-09-10T15:29:12+08:00 Chenguang Li CANN: implement LRU cache for ACL graphs (#15814)
|
||
86587da03bd78df8f4e7d8b111a0c1d2494d6ed0 ff02caf9eed261423289d1531a56536fbf57bfc2 2025-09-10T05:33:58+02:00 Daniel Bevenius llama : check returned fn ptrs from ggml_backend_reg_get_proc_address (#15893)
|
||
ff02caf9eed261423289d1531a56536fbf57bfc2 ae355f6f7108540297d8b7f7ae71d20fe610a0b7 2025-09-10T05:23:19+02:00 Daniel Bevenius ci : cache ROCm installation in windows-latest-cmake-hip (#15887)
|
||
ae355f6f7108540297d8b7f7ae71d20fe610a0b7 4f63cd705c7b6f457f36c63fdc053e07f6f3cc6b 2025-09-09T22:26:03+02:00 Ruben Ortlam vulkan: throw the oom error instead of no memory type found (#15905)
|
||
4f63cd705c7b6f457f36c63fdc053e07f6f3cc6b 17bc5a815f0bebc1844c99796760cd7df5e9b8d9 2025-09-09T07:41:15-05:00 Jeff Bolz vulkan: Fix OOB accesses in soft_max_back (#15861)
|
||
17bc5a815f0bebc1844c99796760cd7df5e9b8d9 ed54e32558ffe0f2c3b1d31f3227211fae05bd49 2025-09-09T14:04:43+02:00 Johannes Gäßler HIP: use v_dot2_f32_f16 instruction for FA (#15884)
|
||
ed54e32558ffe0f2c3b1d31f3227211fae05bd49 a972faebed5fdc4a3d2a844d92d476058c02e02d 2025-09-09T15:01:15+03:00 lksj92hs Workaround for subgroup arithmetic failing on MoltenVK with AMD GPUs (issue 15846) (#15886)
|
||
a972faebed5fdc4a3d2a844d92d476058c02e02d 550cf726e133fd0a069d991287fd3a2a3e3e1cbd 2025-09-09T14:38:02+08:00 Aman Gupta CUDA: Add mul_mat_id support for the mmf kernel (#15767)
|
||
550cf726e133fd0a069d991287fd3a2a3e3e1cbd c252ce67c4b99f056eeb18d42a289e28fee03475 2025-09-09T08:11:01+02:00 Johannes Gäßler CUDA: fix GET_ROWS for large tensors (#15882)
|
||
c252ce67c4b99f056eeb18d42a289e28fee03475 70cd37dbbebdb7a2f84f08207f18eabb0b291a55 2025-09-09T08:42:10+03:00 Georgi Gerganov contrib : add notes about merging PRs (#15881)
|
||
70cd37dbbebdb7a2f84f08207f18eabb0b291a55 acc1b008cfd95e63d5f99a370dbffb98e5a99d2c 2025-09-09T06:06:52+02:00 Daniel Bevenius requirements : update transformers/torch for Embedding Gemma (#15828)
|
||
acc1b008cfd95e63d5f99a370dbffb98e5a99d2c 7057faf64b514e991e2f70147f82bb13d544b1c0 2025-09-09T06:05:55+02:00 Piotr Wilkin (ilintar) model-conversion : add extra debugging support for model conversion (#15877)
|
||
7057faf64b514e991e2f70147f82bb13d544b1c0 fe1c92cd7bb491e9c5767c9de413f2b7dc1e832b 2025-09-08T16:14:32-05:00 Aldehir Rojas json : support `enum` values within `allOf` (#15830)
|
||
fe1c92cd7bb491e9c5767c9de413f2b7dc1e832b e68aa10d8f3d26fdad5b912540362d79de5460e3 2025-09-08T19:57:01+01:00 j-k media : add llama1 icon (#15878)
|
||
e68aa10d8f3d26fdad5b912540362d79de5460e3 0a16bf52e6874369cb4fd2bdc4863ef984b688e7 2025-09-08T13:10:07-05:00 Jeff Bolz vulkan: sort graph to allow more parallel execution (#15850)
|
||
0a16bf52e6874369cb4fd2bdc4863ef984b688e7 88021565f08e0b7c4e07ac089a15ec16fae9166c 2025-09-09T01:23:46+08:00 Aman Gupta CUDA: generate_cu_files.py - add missing mxfp4 (#15880)
|
||
88021565f08e0b7c4e07ac089a15ec16fae9166c 56920f56651908d5cce7a310dabf54ac4f6fbb7f 2025-09-08T10:59:48-04:00 Jesse chat : Deepseek V3.1 reasoning and tool calling support (OpenAI Style) (#15533)
|
||
56920f56651908d5cce7a310dabf54ac4f6fbb7f b0d52998b962bd2681c34bf52af993af79f178b8 2025-09-08T21:50:05+07:00 Xuan-Son Nguyen server : bring back timings_per_token (#15879)
|
||
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)
|
||
f28d4f4ac963f182ea9d0fe9e269f3f5f3782aaf 9fcb29f22f5c33c04c7f0daebb24057899d67a1a 2025-09-08T13:34:56+03:00 Georgi Gerganov metal : refactor + optimize (#15857)
|
||
9fcb29f22f5c33c04c7f0daebb24057899d67a1a 5ef22d281de9c5eaaf616874bc490b89241128cb 2025-09-08T17:33:01+07:00 Xuan-Son Nguyen ggml: allow casting between f32 and i32 (#15783)
|
||
5ef22d281de9c5eaaf616874bc490b89241128cb 233d773d02c37982badddf3994e9953a175f34da 2025-09-08T11:55:44+02:00 Sigbjørn Skjæret CUDA: non-contiguous src0 not supported for PAD (#15869)
|
||
233d773d02c37982badddf3994e9953a175f34da a885dcff11a7b73f9377812d6151f6b15d307de0 2025-09-08T09:44:34+02:00 Daniel Bevenius convert : force setting sliding_window from original config (#15867)
|
||
a885dcff11a7b73f9377812d6151f6b15d307de0 663027fd5490438ce9c27ea866a560e1e268d11f 2025-09-08T10:27:07+03:00 Georgi Gerganov batched-bench : fix llama_synchronize usage during prompt processing (#15835)
|
||
663027fd5490438ce9c27ea866a560e1e268d11f cf0e3ba1500bd23635b444f64ba23cbdb56c92ef 2025-09-08T10:26:36+03:00 Georgi Gerganov context : fix n_outputs during reserve (#15858)
|
||
cf0e3ba1500bd23635b444f64ba23cbdb56c92ef d413dca00360a7e4cb71441dacecfa32556fcc31 2025-09-08T10:25:33+03:00 Georgi Gerganov model : avoid ggml_cont_3d for fused QKV weights (#15662)
|
||
d413dca00360a7e4cb71441dacecfa32556fcc31 85ca66a74676e6d5df4433016488e039a4b464ae 2025-09-07T23:23:41-05:00 Jeff Bolz tests: large sizes for get_rows (#15687)
|
||
85ca66a74676e6d5df4433016488e039a4b464ae 3976dfbe00f02a62c0deca32c46138e4f0ca81d8 2025-09-08T10:03:29+08:00 Chenguang Li CANN: Stream sync between devices for acl_graph (#15809)
|
||
3976dfbe00f02a62c0deca32c46138e4f0ca81d8 d36e61c580bf7fc7879c443c542312a42b718e11 2025-09-07T13:50:26-05:00 Jeff Bolz vulkan: support im2col_3d (#15795)
|
||
d36e61c580bf7fc7879c443c542312a42b718e11 c97b5e5854b47b18a248d77edb693c63018a0865 2025-09-08T02:18:28+08:00 Aaron Teo ggml-cpu: clean up s390x SIMD (#15855)
|
||
c97b5e5854b47b18a248d77edb693c63018a0865 267e99867f09bec8bcc2e424ad9bcddd6cccf9d0 2025-09-07T12:00:49-05:00 Jeff Bolz vulkan: Support pad_ext (#15794)
|
||
267e99867f09bec8bcc2e424ad9bcddd6cccf9d0 3b15924d71237a43bb5ad71f5b885ee66a821342 2025-09-07T11:53:07-05:00 Jeff Bolz vulkan: Use larger loads in scalar/coopmat1 matmul (#15729)
|
||
3b15924d71237a43bb5ad71f5b885ee66a821342 79bc429262268ad2ac8a364cfe6c2d6b9c5f008a 2025-09-07T10:19:45+02:00 Daniel Bevenius ggml WebGPU: remove userdata from request adapter callback (#15527)
|
||
79bc429262268ad2ac8a364cfe6c2d6b9c5f008a c4df49a42d396bdf7344501813e7de53bc9e7bb3 2025-09-07T00:26:28+02:00 Johannes Gäßler CUDA: faster tile FA (Pascal/AMD), headsize 256 (#15769)
|
||
c4df49a42d396bdf7344501813e7de53bc9e7bb3 3c3635d2f20424d557b5b0605a2a356214ffe048 2025-09-06T16:08:43+02:00 Charles Xu kleidiai: generalize compute_forward_kv_cache to compute_forward_fp16 (#15817)
|
||
3c3635d2f20424d557b5b0605a2a356214ffe048 61bdfd5298a78593be649a1035ee2a120b13c4f0 2025-09-06T19:45:24+07:00 Xuan-Son Nguyen server : speed up tests (#15836)
|
||
61bdfd5298a78593be649a1035ee2a120b13c4f0 01806e77714ae8a78130d432945b959a0956c56f 2025-09-06T18:35:04+07:00 Xuan-Son Nguyen server : implement prompt processing progress report in stream mode (#15827)
|
||
01806e77714ae8a78130d432945b959a0956c56f 186415d59552bd5c90549cd8e9be3cc287621fd9 2025-09-06T13:28:44+02:00 Johannes Gäßler ggml-cpu: document use of "free" memory [no ci] (#15834)
|
||
186415d59552bd5c90549cd8e9be3cc287621fd9 fd621880f3fd908424e675a41715a2dc760247a2 2025-09-06T11:27:28+08:00 Aaron Teo ggml-cpu: drop support for nnpa intrinsics (#15821)
|
||
fd621880f3fd908424e675a41715a2dc760247a2 4281c7b315f8a904549fe527039b976e08098d1a 2025-09-05T17:32:39-06:00 Gabe Goodhart aLoRA Support (#15327)
|
||
4281c7b315f8a904549fe527039b976e08098d1a 5fac79cbc77b6d12c9feb5f34fc63586b35fd561 2025-09-06T01:21:15+02:00 Sigbjørn Skjæret ci : exempt correct research label (#15825)
|
||
5fac79cbc77b6d12c9feb5f34fc63586b35fd561 408ff524b40baf4f51a81d42a9828200dd4fcb6b 2025-09-05T14:31:24-06:00 Gabe Goodhart Thinking model disabled assistant prefill (#15404)
|
||
408ff524b40baf4f51a81d42a9828200dd4fcb6b 5143fa895e7725c5bd2135daf7d8f793d98fa91c 2025-09-05T19:43:59+01:00 Eric Curtin Implement --log-colors with always/never/auto (#15792)
|
||
5143fa895e7725c5bd2135daf7d8f793d98fa91c 3a550b5ca4565c9e28f63880d47840feb27d0ff6 2025-09-05T16:07:02+02:00 Johannes Gäßler CUDA: fastdiv, launch bounds for mmvq + q8_1 quant (#15802)
|
||
3a550b5ca4565c9e28f63880d47840feb27d0ff6 a81283820a466f2ace06ce4d4bc9512761f9365f 2025-09-05T14:49:21+02:00 Daniel Bevenius tests : add --list-ops and --show-coverage options (#15745)
|
||
a81283820a466f2ace06ce4d4bc9512761f9365f c610b6c11b1ef7d678671dcf15acd7187a7ad8f3 2025-09-05T11:34:28+02:00 Erik Scholz gguf: gguf_writer refactor (#15691)
|
||
c610b6c11b1ef7d678671dcf15acd7187a7ad8f3 5d6688de08e73acc2532d668380801ed79d704eb 2025-09-05T10:39:22+03:00 Georgi Gerganov kv-cache : fix SWA checks + disable cacheless iSWA (#15811)
|
||
dcbe998a6d8f13b6c9cb36bbe7b355ced2ac3842 87932ec016f0ec81e98a13ce5212851f8c12458a 2025-09-05T14:34:07+08:00 Yunzhe Jia add calrt API used in server side
|
||
5d6688de08e73acc2532d668380801ed79d704eb 4fd1242bef6cb2325b4ff1c1a80f3b54b64508a6 2025-09-05T04:36:23+02:00 Daniel Bevenius model-conversion : add --embeddings flag to modelcard.template [no ci] (#15801)
|
||
4fd1242bef6cb2325b4ff1c1a80f3b54b64508a6 b2426e469e2fdb6c44216d56baa4cfff4f39ae00 2025-09-05T01:24:08+02:00 ExtReMLapin chat : fixed crash when Hermes 2 <tool_call> had a newline before it (#15639)
|
||
b2426e469e2fdb6c44216d56baa4cfff4f39ae00 9e2b1e83c68a38ea0c64f726dd979439bd02189b 2025-09-05T01:22:22+02:00 Piotr Wilkin (ilintar) chat : nemotron thinking & toolcalling support (#15676)
|
||
9e2b1e83c68a38ea0c64f726dd979439bd02189b fb15d649ed14ab447eeab911e0c9d21e35fb243e 2025-09-05T01:05:12+02:00 Piotr Wilkin (ilintar) scripts : add Jinja tester PySide6 simple app (#15756)
|
||
fb15d649ed14ab447eeab911e0c9d21e35fb243e 856ed0947f27b4ec3ad269fceda0402fbab263d3 2025-09-04T18:10:29+02:00 Daniel Bevenius llama : add support for EmbeddingGemma 300m (#15798)
|
||
856ed0947f27b4ec3ad269fceda0402fbab263d3 d1e2adba65208d6de3d7f6f23b00c562ff5bb777 2025-09-04T09:53:22-06:00 Gabe Goodhart metal : Add template specialization for mul_mm_id w/ ne20 == 10 (#15799)
|
||
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)
|
||
a68d9144262f1d0ef4f6ba7ad4a7e73e977ba78c badb80cadbc40e047b30c43611aba575fc8d6845 2025-09-04T11:50:23+02:00 Xuan-Son Nguyen server: add exceed_context_size_error type (#15780)
|
||
badb80cadbc40e047b30c43611aba575fc8d6845 0a1b3982cd0bd18730d50a693053b88c13fd04a6 2025-09-04T10:49:44+01:00 Eric Curtin Document the new max GPU layers default in help (#15771)
|
||
0a1b3982cd0bd18730d50a693053b88c13fd04a6 5421f63ab08ca1e1a093662a5ccd0117e461185f 2025-09-04T16:38:49+08:00 leejet ggml: add ops for WAN video model (cuda && cpu) (#15669)
|
||
5421f63ab08ca1e1a093662a5ccd0117e461185f 820bc9853100708011036e10f40b923968dd9b66 2025-09-04T15:12:30+08:00 hipudding CANN: Fix precision issue on 310I DUO multi-devices (#15784)
|
||
820bc9853100708011036e10f40b923968dd9b66 239b60e8986bbcb944f57311a02ac994431bf652 2025-09-04T08:30:28+02:00 rmatif opencl: add hs=40 to FA (#15758)
|
||
239b60e8986bbcb944f57311a02ac994431bf652 dff7551bfdbdd6e57c13e523d7dcca317640e907 2025-09-04T11:03:02+08:00 Chenguang Li CANN: fix acl_rstd allocation size in ggml_cann_rms_norm (#15760)
|
||
dff7551bfdbdd6e57c13e523d7dcca317640e907 0fce7a1248b74148c1eb0d368b7e18e8bcb96809 2025-09-03T22:55:10+02:00 Ruben Ortlam vulkan: fix mmv subgroup16 selection (#15775)
|
||
0fce7a1248b74148c1eb0d368b7e18e8bcb96809 8227695d7a3e5b357cb37fad263f00a8ca6db710 2025-09-03T13:33:15-05:00 Jeff Bolz vulkan: don't use std::string in load_shaders, to improve compile time (#15724)
|
||
8227695d7a3e5b357cb37fad263f00a8ca6db710 0014fb4add3a7d000c14b763538045c6170f57d9 2025-09-03T20:24:50+02:00 Daniel Bevenius vulkan : update ggml_vk_instance_validation_ext_available (#15666)
|
||
0014fb4add3a7d000c14b763538045c6170f57d9 661ae31c9c68201577e70278285b349a5a662caf 2025-09-04T03:22:55+09:00 Shin-myoung-serp ggml vulkan: add hardsigmoid and hardswish operations (#15762)
|
||
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)
|
||
407c23786dd0d3a503e9429eead96e611d3950c9 cdedb70a998cea7052560fe0b0615a839443564d 2025-09-03T18:28:36+02:00 Daniel Bevenius model-conversion : fix pyright errors (#15770)
|
||
cdedb70a998cea7052560fe0b0615a839443564d 2c8dac72eb6acd4e20c0da251535dfc46d35178b 2025-09-03T18:16:26+03:00 Georgi Gerganov sampling : optimize dist sampler (#15704)
|
||
2c8dac72eb6acd4e20c0da251535dfc46d35178b 40a751ea9a94364da73537b86502a808ebe1fc3a 2025-09-03T13:35:49+02:00 Daniel Bevenius llama : fix incorrect model type for Gemma 270M (#15764)
|
||
40a751ea9a94364da73537b86502a808ebe1fc3a 5eae9348835037046f33fafd852df7c952c95209 2025-09-03T12:50:47+02:00 Daniel Bevenius model-conversion : remove hardcoded /bin/bash shebangs [no ci] (#15765)
|
||
5eae9348835037046f33fafd852df7c952c95209 05c0380f2ae5c7193445e03ebc7b6de9ce49f1a6 2025-09-03T16:46:01+08:00 hipudding CANN: Add RoPE contiguous check for 310I DUP device (#15735)
|
||
05c0380f2ae5c7193445e03ebc7b6de9ce49f1a6 8c3fdf44ecf08335942f2ba558955f55e88c7991 2025-09-03T16:16:21+08:00 xctan ggml-cpu : optimize RVV kernels (#15720)
|
||
8c3fdf44ecf08335942f2ba558955f55e88c7991 f6da8cb86a28f0319b40d9d2a957a26a7d875f8c 2025-09-03T09:48:35+02:00 Daniel Bevenius model-conversion : add missing curl script [no ci] (#15761)
|
||
f6da8cb86a28f0319b40d9d2a957a26a7d875f8c 8a2234ea0c89f212190c176d741b7742f0082582 2025-09-03T14:08:22+08:00 hipudding CANN: Mask unsupported TRANSPOSE_1D operator (#15733)
|
||
8a2234ea0c89f212190c176d741b7742f0082582 3de008208b9b8a33f49f979097a99b4d59e6e521 2025-09-03T10:43:53+08:00 Chenguang Li CANN: Fix type float_t to float (#15736)
|
||
3de008208b9b8a33f49f979097a99b4d59e6e521 69db8a52e6dd0db81ec05c581150cc1e43b8ac46 2025-09-03T03:27:30+08:00 SnA1lGo fix: resolve unsigned int initialization warning for n_dims/size in gguf.cpp (#15754)
|
||
69db8a52e6dd0db81ec05c581150cc1e43b8ac46 c466abe1587bde458c806e2134e37750f2edf8fb 2025-09-02T19:40:37+02:00 Oliver Simons chore: Update `.clang-format` to use `BinPackArguments=true` (#15744)
|
||
c466abe1587bde458c806e2134e37750f2edf8fb 0a2a3841e8ebc570da343e8cf58a21c1010b41e7 2025-09-02T18:17:26+02:00 Johannes Gäßler llama: -fa 1/0/-1 aliases for -fa on/off/auto (#15746)
|
||
0a2a3841e8ebc570da343e8cf58a21c1010b41e7 9961d244f2df6baf40af2f1ddc0927f8d91578c8 2025-09-02T16:02:26+02:00 Ruben Ortlam vulkan: fix shaders gen when no integer dot is available (#15740)
|
||
9961d244f2df6baf40af2f1ddc0927f8d91578c8 25f1045f07cf0daf667d63e35618842e3174a8c7 2025-09-02T17:12:37+08:00 hipudding CANN: Resolve soft_max precision issue (#15730)
|
||
25f1045f07cf0daf667d63e35618842e3174a8c7 97669e40735d08db65ef094a8faae8e8411a97db 2025-09-02T01:37:01-05:00 Jeff Bolz vulkan: Fix macro parameter order for f32 matmul shaders (#15716)
|
||
97669e40735d08db65ef094a8faae8e8411a97db 2f853687b3bce15e143a22f678d1715060fd606c 2025-09-02T08:26:53+02:00 rmatif opencl: add attn sinks support for FA kernels (#15706)
|
||
2f853687b3bce15e143a22f678d1715060fd606c ef2af57ddf04ab367f6ba6e8ed100fc56785e4b7 2025-09-02T14:07:48+08:00 Chenguang Li CANN: Support eager execution mode under ACL graph compilation (#15712)
|
||
ef2af57ddf04ab367f6ba6e8ed100fc56785e4b7 5d804a4938d896f9089687a8b99911cdb43b0b94 2025-09-02T14:05:23+08:00 hipudding CANN: Support ext_factor in rope (#15710)
|
||
5d804a4938d896f9089687a8b99911cdb43b0b94 d4d8dbe383e8b9600cbe8b42016e3a4529b51219 2025-09-02T01:14:55+02:00 Johannes Gäßler ggml-backend: raise GGML_MAX_SPLIT_INPUTS (#15722)
|
||
d4d8dbe383e8b9600cbe8b42016e3a4529b51219 35a42edac84690990a75726898d975cd08d220c0 2025-09-01T22:17:42+03:00 Gilad S. vulkan: use memory budget extension to read memory usage (#15545)
|
||
35a42edac84690990a75726898d975cd08d220c0 fec7911f8febb22c27ce1acb12441800da24cc06 2025-09-01T14:01:10-05:00 Jeff Bolz vulkan: add missing clamps in new mul_mat_id paths (#15702)
|
||
fec7911f8febb22c27ce1acb12441800da24cc06 078ce23ea77988f2fe1a42afb18d58bc084d55fa 2025-09-01T20:58:35+02:00 Ruben Ortlam vulkan: disable large mmv subgroups on older Nvidia GPUs (#15717)
|
||
078ce23ea77988f2fe1a42afb18d58bc084d55fa a0c2b207c596d1092a08615de61ab56a7d63515f 2025-09-02T03:13:49+09:00 s-goto-11 ggml: SVE support for exponential functions (#15145)
|
||
a0c2b207c596d1092a08615de61ab56a7d63515f 4b20d8b7e31718d3fe8b4f12d220a9d19e4f1997 2025-09-01T23:43:16+05:30 Prashant Vithule ggml: aarch64: Implement SVE F16 kernels for vector functions (#15115)
|
||
4b20d8b7e31718d3fe8b4f12d220a9d19e4f1997 02c1813517412f3e00aa6ca7c0273fea64edb492 2025-09-01T23:53:31+08:00 Jie Fu (傅杰) convert : remove redundant code (#15708)
|
||
02c1813517412f3e00aa6ca7c0273fea64edb492 77dee9de97be75b7143a213bc48893e0c0b29af7 2025-09-01T16:19:07+02:00 Ruben Ortlam Vulkan: Add Integer Dot Product mul_mat_vec shader for legacy quants (#14903)
|
||
77dee9de97be75b7143a213bc48893e0c0b29af7 4795c91c32fec7165a1364763d4d4f0c93abf933 2025-09-01T14:28:49+02:00 Daniel Bevenius ggml : WebGPU add TRANSPOSE and RESHAPE to supported ops (#15695)
|
||
4795c91c32fec7165a1364763d4d4f0c93abf933 b66df9d9c942254d03209186ef24ed7c994a576e 2025-09-01T15:34:59+08:00 Jie Fu (傅杰) docs : add Hunyuan to models section (#15707)
|
||
b66df9d9c942254d03209186ef24ed7c994a576e b9382c3877c6067feccf182efe9449a2d1cb24c7 2025-09-01T06:55:06+05:30 Akarshan Biswas CUDA: fix build error from ambiguous __half conversions in conv2d (#15690)
|
||
b9382c3877c6067feccf182efe9449a2d1cb24c7 3dc7397a2799bdc07bccf637ab7ae5a1e786d1a4 2025-09-01T08:57:23+08:00 hipudding CANN: Optimize MUL_MAT_ID (#15658)
|
||
3dc7397a2799bdc07bccf637ab7ae5a1e786d1a4 e92d53b29e393fc4c0f9f1f7c3fe651be8d36faa 2025-09-01T08:57:00+08:00 hipudding CANN: fix RoPE cache issue on multi-device (#15629)
|
||
e92d53b29e393fc4c0f9f1f7c3fe651be8d36faa 0d161f021aa33ec0e90cce96f5d1a88925557327 2025-08-31T20:41:02+03:00 Georgi Gerganov sampling : optimize samplers by reusing bucket sort (#15665)
|
||
0d161f021aa33ec0e90cce96f5d1a88925557327 4efd5a83163ff383285b3a4c2106feabf5c69557 2025-08-31T20:11:58+03:00 Georgi Gerganov server : enable /slots by default and make it secure (#15630)
|
||
4efd5a83163ff383285b3a4c2106feabf5c69557 274966226f87f301ac132da898280ca3142b60e5 2025-08-31T19:43:30+03:00 Georgi Gerganov metal : fix checks for available FA kernels (#15700)
|
||
274966226f87f301ac132da898280ca3142b60e5 9777032dccd67bdc7785aeab7497014a8be8dacc 2025-08-31T08:47:05-07:00 Diego Devesa llama : fix fattn reserve call n_seqs parameter (#15699)
|
||
9777032dccd67bdc7785aeab7497014a8be8dacc 7d3c9f2b217acf0ce5db81ae83d3f375f49ab2c7 2025-08-31T06:49:03-07:00 Diego Devesa llama : separate compute buffer reserve from fattn check (#15696)
|
||
7d3c9f2b217acf0ce5db81ae83d3f375f49ab2c7 bbbf5ecccb35286521f735239d499eec4279a840 2025-08-31T15:30:20+02:00 Sigbjørn Skjæret ci : explicitly set fa off or on (#15692)
|
||
bbbf5ecccb35286521f735239d499eec4279a840 c37052ab4d6d1ae73c0e90bc6e560cc6409e1311 2025-08-31T03:13:27-05:00 Jeff Bolz vulkan: handle large sizes for get_rows (#15686)
|
||
c37052ab4d6d1ae73c0e90bc6e560cc6409e1311 5c16b9c87d840e4d5d55fa83c732c6b693346f40 2025-08-31T02:06:43-05:00 Jeff Bolz vulkan: mul_mat_id coopmat2 optimizations (#15546)
|
||
5c16b9c87d840e4d5d55fa83c732c6b693346f40 b97c9edc59d4a1b4069991aa670411190f4f3a3e 2025-08-31T08:46:42+02:00 Daniel Bevenius vulkan : remove unused portability_enumeration_ext variable (#15679)
|
||
b97c9edc59d4a1b4069991aa670411190f4f3a3e 94e82c7eadeb8fff0db4bfd1ab6d8cf65fa6f2e0 2025-08-31T01:30:54-05:00 Jeff Bolz vulkan: Allow fallback to sysmem memory when vidmem is full (#15649)
|
||
94e82c7eadeb8fff0db4bfd1ab6d8cf65fa6f2e0 4d74393bcc956ccd7df68a6a06d1a0575cfa712c 2025-08-31T01:27:57-05:00 Jeff Bolz vulkan: clamp matmul and FA results to the max finite value (#15652)
|
||
4d74393bcc956ccd7df68a6a06d1a0575cfa712c dd892555b0681b7f56d38780f6fdfe00a195160f 2025-08-30T18:03:42+02:00 Charles Xu ggml: update kleidiai to v1.13.0 (#15663)
|
||
dd892555b0681b7f56d38780f6fdfe00a195160f e81b8e4b7f5ab870836fad26d154a7507b341b36 2025-08-30T08:51:28-07:00 Diego Devesa Update build.md to remove MSVC arm64 notes (#15684)
|
||
e81b8e4b7f5ab870836fad26d154a7507b341b36 38ad381f9f5d4dd368a96d844fb19cf501ed9d22 2025-08-30T16:32:10+02:00 Johannes Gäßler llama: use FA + max. GPU layers by default (#15434)
|
||
38ad381f9f5d4dd368a96d844fb19cf501ed9d22 696fccf354e9dbdfbce135bc40b44c9dcc64dda9 2025-08-30T16:20:32+02:00 Johannes Gäßler CUDA: use FP32 arithmetic for conv2d (#15683)
|
||
696fccf354e9dbdfbce135bc40b44c9dcc64dda9 ef476916bba4b44f44be0c98babc1cb025968e75 2025-08-30T04:11:22-05:00 Jeff Bolz vulkan: Skip syncing for prealloc_y when it is reused (#15544)
|
||
ef476916bba4b44f44be0c98babc1cb025968e75 d82f6aa34a216f5df1945cdfe121ba5e6cd80be0 2025-08-30T10:18:35+08:00 Chenguang Li CANN: FIx compiler warnings (#15661)
|
||
d82f6aa34a216f5df1945cdfe121ba5e6cd80be0 3d16b29c3bb1ec816ac0e782f20d169097063919 2025-08-30T00:12:53+02:00 Sergey Alirzaev server : removed obsolete doc (#15670)
|
||
3d16b29c3bb1ec816ac0e782f20d169097063919 792b44f2ed9668cce7f267ff0ae4950ed9b4a5de 2025-08-29T22:04:08+02:00 Johannes Gäßler scripts: strip "AMD Instinct" from GPU name (#15668)
|
||
792b44f2ed9668cce7f267ff0ae4950ed9b4a5de 81017865ee444cf49ce0136f2be1e41a0270ff91 2025-08-29T19:25:40+02:00 ExtReMLapin server : add documentation for `parallel_tool_calls` param (#15647)
|
||
81017865ee444cf49ce0136f2be1e41a0270ff91 60e5eee31f1af9bb579ac45380e3857d610020b9 2025-08-29T21:30:06+08:00 Aman Gupta CUDA: fix bug in rms_norm fusion (#15660)
|
||
60e5eee31f1af9bb579ac45380e3857d610020b9 009b709d6efd24820ac67765ed339a72dc797814 2025-08-29T14:53:41+02:00 Piotr Wilkin (ilintar) chat : Seed OSS thinking + tool call support (#15552)
|
||
009b709d6efd24820ac67765ed339a72dc797814 e8d99dd0b67f2ecc1e45fca8074a3a18c3e036d2 2025-08-29T11:35:58+08:00 Aman Gupta CUDA: fuse adds, fuse add with rms norm (#15631)
|
||
e8d99dd0b67f2ecc1e45fca8074a3a18c3e036d2 a8bca68f727844e7dcf24a956003b3c2039ea563 2025-08-28T18:39:31-06:00 Gabe Goodhart nvidia nemotron nano v2 (nemotronh) (#15507)
|
||
a8bca68f727844e7dcf24a956003b3c2039ea563 c97dc093912ad014f6d22743ede0d4d7fd82365a 2025-08-28T15:27:36-05:00 Gabe Goodhart fix: Compute the full sum in llama-eval-callback, not just the sum of printed values (#15637)
|
||
c97dc093912ad014f6d22743ede0d4d7fd82365a 6c442f42ff25564a0cd6b1435d9abc1b0178eac5 2025-08-29T00:03:03+05:30 mnehete32 CUDA: add conv2d (#15635)
|
||
6c442f42ff25564a0cd6b1435d9abc1b0178eac5 73804145ab6c06888e314046abf08f66a0133679 2025-08-28T22:39:27+08:00 Aaron Teo ggml-cpu: fix invalid hsum build in debug s390x (#15634)
|
||
73804145ab6c06888e314046abf08f66a0133679 c8d0d14e77c3c45df5cbbddde9c2b4c377ec4e7a 2025-08-28T10:11:36-04:00 compilade ggml : fix SSM_SCAN for n_groups > 1 (#15625)
|
||
c8d0d14e77c3c45df5cbbddde9c2b4c377ec4e7a 84ab83cc0b4b7e769451ee48e4c7d1acef91ef25 2025-08-28T17:09:05+03:00 Georgi Gerganov kv-cache : fix find_slot to not search for continuous slot (#15638)
|
||
84ab83cc0b4b7e769451ee48e4c7d1acef91ef25 55042b3692cb1467c9ee15c62c4a9fbf180f89e3 2025-08-28T15:49:50+02:00 Sigbjørn Skjæret model : jina-embeddings-v3 support (#13693)
|
||
55042b3692cb1467c9ee15c62c4a9fbf180f89e3 8a4280ce431da6b33e5a95ae1fd61472c8c3f8cc 2025-08-28T19:23:22+08:00 Aman Gupta scripts: add sqlite3 check for compare-commits.sh (#15633)
|
||
8a4280ce431da6b33e5a95ae1fd61472c8c3f8cc 64387f6e95434b393ac3df285864692b7fd9c4d2 2025-08-28T12:27:02+03:00 Georgi Gerganov kv-cache : remove LLAMA_SET_ROWS checks (#15505)
|
||
64387f6e95434b393ac3df285864692b7fd9c4d2 d35a1e8c41f747548775225973a99507896a8c61 2025-08-28T10:56:41+02:00 Aleksei Nikiforov gguf-py: byteswapping improvements (#12851)
|
||
d35a1e8c41f747548775225973a99507896a8c61 46d9caa27a0281150e8cf082308c0f9e7576ebe5 2025-08-28T01:48:20-06:00 Joshua Cogliati cli : change log to warning to explain reason for stopping (#15604)
|
||
46d9caa27a0281150e8cf082308c0f9e7576ebe5 5a0e3ef6f00c658fbae53797f02d5a360ebf8fec 2025-08-28T09:26:48+02:00 Daniel Bevenius model-conversion : add mmproj conversion target (#15628)
|
||
5a0e3ef6f00c658fbae53797f02d5a360ebf8fec fbef0fad7a7c765939f6c9e322fa05cd52cf0c15 2025-08-27T17:32:36-07:00 matiaslin cuda: Add cublasLt_static linking when GGML_STATIC is enabled (#15622)
|
||
fbef0fad7a7c765939f6c9e322fa05cd52cf0c15 da54f9f1a2db07aaae390024ac466e7867685d94 2025-08-27T20:58:09+02:00 Johannes Gäßler server: higher timeout for tests (#15621)
|
||
da54f9f1a2db07aaae390024ac466e7867685d94 47373271f971aa5a0a6462b286f4a7b5bd4ba644 2025-08-27T15:48:07+03:00 Georgi Gerganov presets : add qwen3-30B-a3b FIM (#15616)
|
||
47373271f971aa5a0a6462b286f4a7b5bd4ba644 1bded5a3b3376b2aa3ba2f11ade5910c550f354b 2025-08-27T13:58:54+02:00 uvos HIP: Enable support for ggml_backend_cuda_register_host_buffer (#15615)
|
||
1bded5a3b3376b2aa3ba2f11ade5910c550f354b 1e7489745a74996fc36e8fd05b73aa16bc184e0c 2025-08-27T13:55:12+03:00 Georgi Gerganov kv-cache : better estimate of n_kv for multi-sequence batches (#15610)
|
||
1e7489745a74996fc36e8fd05b73aa16bc184e0c 1cf123a343ab7ca5586aacb9e0a1d2de7fe33be4 2025-08-27T17:21:41+08:00 Chenguang Li CANN: refactor mask handling and improve performance in FA (#15561)
|
||
1cf123a343ab7ca5586aacb9e0a1d2de7fe33be4 fcca2182a18c786c57d02d1d1927204b133f1fdc 2025-08-27T16:44:22+08:00 xctan ggml-cpu : add basic RVV support for vector f32 ops (#15057)
|
||
fcca2182a18c786c57d02d1d1927204b133f1fdc 86076f92de1a547ef87c304facca3b4b9fae6c21 2025-08-27T10:28:53+02:00 Daniel Bevenius common : add -m to bash completion for --model [no ci] (#15591)
|
||
86076f92de1a547ef87c304facca3b4b9fae6c21 bcbddcd54f0d5c22eab180831fdea6484107112f 2025-08-27T08:36:05+02:00 rmatif OpenCL: add fused group_norm/norm, mul, add (#15314)
|
||
87932ec016f0ec81e98a13ce5212851f8c12458a 0115cfb7c1dec0753aecb34f3af4338bf9de8eeb 2025-08-27T14:29:18+08:00 Yunzhe Jia fix merge problem
|
||
bcbddcd54f0d5c22eab180831fdea6484107112f 8b696861364360770e9f61a3422d32941a477824 2025-08-26T13:14:38-07:00 Diego Devesa tests : fix test-opt with GGML_BACKEND_DL (#15599)
|
||
8b696861364360770e9f61a3422d32941a477824 8ce3ff1d91245e158d98d8062cd64b0dd98dcfe3 2025-08-27T00:27:49+05:30 Akarshan Biswas SYCL: fix rms_norm_mul_add for tensor dim not a multiple of sg_size (#15592)
|
||
8ce3ff1d91245e158d98d8062cd64b0dd98dcfe3 44b1efa41acd4df3f56ee0e46f898135ecd1a054 2025-08-26T20:05:50+02:00 fidoriel mtmd : fix mtmd ios build (#15579)
|
||
44b1efa41acd4df3f56ee0e46f898135ecd1a054 a6a58d64785cb458ed9de52f391aa38142d38d64 2025-08-26T15:42:49Z Eve tests: add performance test for mul mat id (#15543)
|
||
a6a58d64785cb458ed9de52f391aa38142d38d64 0373486dbc0dccbdcb3b5fdd65759d88cec06196 2025-08-26T21:05:25+05:30 shalinib-ibm llamafile: PowerPC Sgemm Optimization (#15558)
|
||
0373486dbc0dccbdcb3b5fdd65759d88cec06196 62cef26ac5b6b7acb635d3dc963813b43952dc2b 2025-08-26T17:45:17+03:00 Georgi Gerganov graph : fix assert in memory-less build_attn (#15590)
|
||
62cef26ac5b6b7acb635d3dc963813b43952dc2b 8f5afa94c4f929da71f560db7c9f38ef6a783d95 2025-08-26T16:12:29+02:00 Daniel Bevenius model-conversion : add qat-q4 quantization targets (#15588)
|
||
8f5afa94c4f929da71f560db7c9f38ef6a783d95 b3964c1e890ef8c947afb36a5124ce6fcb2136d4 2025-08-26T16:01:20+02:00 Johannes Gäßler CUDA: return -1 for nonexistent compiled arch (#15587)
|
||
b3964c1e890ef8c947afb36a5124ce6fcb2136d4 79a546220c719e6a70627b243a478ab8d84dc9e1 2025-08-26T14:22:14+03:00 Georgi Gerganov metal : optimize FA vec for large sequences and BS <= 8 (#15566)
|
||
79a546220c719e6a70627b243a478ab8d84dc9e1 85cc1ae998e4898d9fa992cb9b8620338cee97bf 2025-08-26T12:54:19+02:00 Xuan-Son Nguyen mtmd : support Kimi VL model (#15458)
|
||
85cc1ae998e4898d9fa992cb9b8620338cee97bf 1d8d83deaa48d4a5491820d58ff1c0d8cf9d196c 2025-08-26T12:47:00+03:00 Georgi Gerganov context : print graph stats for memory-less contexts (#15586)
|
||
1d8d83deaa48d4a5491820d58ff1c0d8cf9d196c c4e9239064a564de7b94ee2b401ae907235a8fca 2025-08-26T12:46:15+03:00 Georgi Gerganov metal : improve `MUL_MAT_ID` (#15541)
|
||
c4e9239064a564de7b94ee2b401ae907235a8fca 39842a7f73012eb42816ca4f26411782bd3da7c5 2025-08-26T16:05:55+08:00 tc-mb model : support MiniCPM-V 4.5 (#15575)
|
||
0115cfb7c1dec0753aecb34f3af4338bf9de8eeb 3a617cb18121a85eed17ac5585788810c5d9c1e0 0d5a470223fc90b6b6807921d68011ff06ae7f9e 2025-08-26T15:19:56+08:00 Yunzhe Jia Merge branch 'master' into dev
|
||
39842a7f73012eb42816ca4f26411782bd3da7c5 0fd90db5858e325358a5fbdaa4e327ee2a79d0d4 2025-08-26T09:08:08+02:00 Sigbjørn Skjæret gguf-py : remove erroneous FFN_GATE entry (#15583)
|
||
0fd90db5858e325358a5fbdaa4e327ee2a79d0d4 4c37636b3ea96f2574eeb7668b93fcc0e64b05dd 2025-08-26T08:51:43+02:00 Sigbjørn Skjæret metal : remove contiguous assertion for src0 in IM2COL (#15577)
|
||
4c37636b3ea96f2574eeb7668b93fcc0e64b05dd 34bdbbd7c2b70b848718e95bc48010f6aecd2816 2025-08-26T13:15:33+07:00 Yoshi_likes_e4 Add a warning for special devices (#15563)
|
||
34bdbbd7c2b70b848718e95bc48010f6aecd2816 74f52f77f28a5ad6d6075231afcb8d1ad763ca32 2025-08-25T23:42:44-05:00 Jeff Bolz vulkan: Remove splitting for mul_mat_id (#15568)
|
||
74f52f77f28a5ad6d6075231afcb8d1ad763ca32 f7207b0415986dd7f48447149da7de3a82338276 2025-08-26T05:21:22+08:00 Qeeweew CUDA: Accelerate MXFP4 table lookup using `__byte_perm` (#15451)
|
||
f7207b0415986dd7f48447149da7de3a82338276 4d917cd4f64cc37744e76d084659475819fb0728 2025-08-25T14:18:09-07:00 lhez opencl: fix support ops condition for `rms_norm` (#15560)
|
||
4d917cd4f64cc37744e76d084659475819fb0728 886b97a5d693550c2da470c091d9d27bf38398f8 2025-08-25T17:56:59+02:00 Ruben Ortlam vulkan: fix min subgroup 16 condition for mmid subgroup optimization (#15565)
|
||
886b97a5d693550c2da470c091d9d27bf38398f8 111f8d06f0b9169059779a35f655b343242ccbb6 2025-08-25T10:47:16-05:00 Jeff Bolz tests: Generate unique input values for count_equal (#15487)
|
||
111f8d06f0b9169059779a35f655b343242ccbb6 5eff6ec9b1220b599a43b594b1110487ab6aca08 2025-08-25T11:27:34-04:00 Ihar Hrachyshka metal: fix regression when no metal devices are present (#15531)
|
||
5eff6ec9b1220b599a43b594b1110487ab6aca08 dfd9b5f6c7586c88588f06a644c131bec071a0a1 2025-08-25T17:23:40+02:00 Johannes Gäßler CUDA: MoE helper in device code, better tile sizes (#15525)
|
||
dfd9b5f6c7586c88588f06a644c131bec071a0a1 5a6bc6b1a6cb665a944426c2055794950e524bf5 2025-08-25T15:00:43+02:00 Daniel Bevenius model-conversion : set pooling type to none in logits.cpp (#15564)
|
||
5a6bc6b1a6cb665a944426c2055794950e524bf5 6b64f74b55628e4193f4fb00313f07dbd8556528 2025-08-25T14:25:25+02:00 Daniel Bevenius model-conversion : add model card template for embeddings [no ci] (#15557)
|
||
6b64f74b55628e4193f4fb00313f07dbd8556528 0d5a470223fc90b6b6807921d68011ff06ae7f9e 2025-08-25T13:56:43+03:00 Georgi Gerganov batched-bench : fix unified KV cache handling + pp timing (#15562)
|
||
3a617cb18121a85eed17ac5585788810c5d9c1e0 c42293848dd9f03eba9a6c1b85a600b398f06541 2025-08-13T10:58:59+08:00 Yunzhe Jia add calrt_decode
|
||
c42293848dd9f03eba9a6c1b85a600b398f06541 1cbf442f50a154aac0bf35d7b8b3436a2cdfd6f5 2025-08-06T15:46:06+08:00 Yunzhe Jia sync with calrt API
|
||
1cbf442f50a154aac0bf35d7b8b3436a2cdfd6f5 ba67d6c03fed3193987a4ef13050e6e2a4451df4 2025-07-24T15:50:19+08:00 Yunzhe Jia 1.sync with calrt prefill,decode,kv_update op 2.add hw pattern untile function
|
||
ba67d6c03fed3193987a4ef13050e6e2a4451df4 8f52c6f8abb2ed9ad967cc0bfb90acb073011193 2025-06-26T17:42:32+08:00 Yunzhe Jia llama-calrt: infer-op: copy data from output buffer to dst-tensor.data
|
||
8f52c6f8abb2ed9ad967cc0bfb90acb073011193 d2f325130c4ed77a3dcff2c23a926587fb08e2cc 2025-06-26T15:10:45+08:00 Yunzhe Jia Squashed commit of the following:
|
||
d2f325130c4ed77a3dcff2c23a926587fb08e2cc 6391ee7ffd97e723dfbdcc32f001e1b95a63cba6 2025-06-17T11:53:33+08:00 Yunzhe Jia calrt: add input_token in graph; input_token will be copy to calrt_in_buffer
|
||
6391ee7ffd97e723dfbdcc32f001e1b95a63cba6 b9850050a5b5147cb2e6720b98ed8c1e8e02eff9 2025-06-13T10:58:06+08:00 Yunzhe Jia update .gitignore to update cal_test.gguf
|
||
b9850050a5b5147cb2e6720b98ed8c1e8e02eff9 933b205c32551848589b8339437d778cb4f2a46d 2025-06-06T17:25:52+08:00 Yunzhe Jia calrt: add call chain for using calrt::infer
|
||
933b205c32551848589b8339437d778cb4f2a46d 33fea9a8592fd338abe279b89350a3af2e0ff1f7 2025-06-12T17:47:19+08:00 Yunzhe Jia gguf-py: add cal_test.py
|
||
33fea9a8592fd338abe279b89350a3af2e0ff1f7 806f05a353d322e97d97e02f910de2b408a8b4ab 2025-06-06T17:25:52+08:00 Yunzhe Jia calrt: add load calbin
|
||
806f05a353d322e97d97e02f910de2b408a8b4ab ec9e0301fef6476df83e94842c3b625501c95566 2025-05-30T09:45:37+08:00 Yunzhe Jia add calrt demo
|