Commit Graph

  • ece20f1d64 Remove tree_map import Angelos Katharopoulos 2024-09-12 16:05:46 -0700
  • b0a42d091e Flush the messages Angelos Katharopoulos 2024-09-12 15:04:21 -0700
  • e0f18d15aa Use concatenated all reduce and gather stats Angelos Katharopoulos 2024-09-12 13:33:57 -0700
  • 4786b4e3eb Add distributed option for lora training Angelos Katharopoulos 2024-05-31 18:42:22 -0700
  • 79075b7a21 fix tests Alex Barron 2024-10-31 12:37:15 -0700
  • 8444ff0f6a in place Alex Barron 2024-10-31 12:22:36 -0700
  • 15dcebc36a use safetensors in whisper Awni Hannun 2024-10-04 10:59:01 -0700
  • 1d53354b51 fix sed Alex Barron 2024-10-31 12:06:34 -0700
  • 2e0690374e single sdpa function Alex Barron 2024-10-31 12:02:34 -0700
  • 29f21e7fe4 add tests Alex Barron 2024-10-28 22:14:52 -0700
  • 37a3723823 simplify Alex Barron 2024-10-28 16:03:43 -0700
  • 48655a7f83 add QuantizedKVCache Alex Barron 2024-10-26 00:23:46 -0700
  • 9900d2230a fix Awni Hannun 2024-10-31 11:57:13 -0700
  • 9f34fdbda4
    Wire models in MLX LM (#1069) Awni Hannun 2024-10-31 08:17:14 -0700
  • 7e46988a1c bump to 0.19.2 Awni Hannun 2024-10-31 07:49:08 -0700
  • e6d35301bd smol modification N8 2024-10-31 02:37:14 -0400
  • 7e4413b1dd add max token limit N8 2024-10-31 02:20:55 -0400
  • 58b448dc0b updates Goekdeniz-Guelmez 2024-10-30 21:23:13 +0100
  • ffc7ab06a0
    Merge branch 'ml-explore:main' into adding-support-for-mamba2 Gökdeniz Gülmez 2024-10-30 17:04:38 +0100
  • e4c8bf1199 mlx lm version Awni Hannun 2024-10-29 17:39:59 -0700
  • f46c9bb461 version Awni Hannun 2024-10-29 17:39:29 -0700
  • 0793fd070e comment + nit Awni Hannun 2024-10-29 17:38:53 -0700
  • 8fe9539af7
    Fix detokenizer space match for quote (#1072) Awni Hannun 2024-10-27 15:06:07 -0700
  • 49813f524f Add support for falcon-mamba Ilyas Chahed 2024-10-27 13:28:49 +0000
  • 6d4290928f format Awni Hannun 2024-10-26 18:14:56 -0700
  • db7d3c7a76 remove transformer flax/torch warning Awni Hannun 2024-10-26 18:11:14 -0700
  • 00b67e2f97 fix + test Awni Hannun 2024-10-26 17:41:55 -0700
  • ab4bf05c6e
    Update lora_config.yaml with new param: num_layers (#1068) hschaeufler 2024-10-26 19:34:46 +0300
  • 67607a8e13 Start memory-efficient flux finetuning branch flux-qlora Angelos Katharopoulos 2024-10-25 09:46:47 -0700
  • dd4009215a
    Merge branch 'ml-explore:main' into main chenguangjian 2024-10-25 16:38:30 +0800
  • 3b70708201
    Merge branch 'ml-explore:main' into adding-support-for-mamba2 Gökdeniz Gülmez 2024-10-25 08:57:37 +0200
  • 4971462bf0
    feat(clip): add linear probe evaluation script (#960) Saurav Maheshkar 2024-10-25 05:56:17 +0100
  • 6d13a14757 Fix linear probe script Angelos Katharopoulos 2024-10-24 21:53:00 -0700
  • 7c8849e795 update Goekdeniz-Guelmez 2024-10-24 16:16:42 +0200
  • 2822acc98a fix synch Awni Hannun 2024-10-24 06:19:19 -0700
  • a039f11b11
    Merge branch 'ml-explore:main' into main 锦此 2024-10-24 15:25:53 +0800
  • 8f8c118419
    Merge pull request #1 from jincdream/add-timeout 锦此 2024-10-24 15:25:34 +0800
  • 131ccbe6df wired in MLX LM Awni Hannun 2024-10-23 19:59:28 -0700
  • ed7333929f
    Merge branch 'main' into feat/batch_generate L Lllvvuu 2024-10-23 18:16:44 -0700
  • 9284e50c67
    Update lora_config.yaml with new param: num_layers hschaeufler 2024-10-23 22:20:35 +0300
  • 9000e280ae
    fix mamba models conversion (#1065) Awni Hannun 2024-10-22 15:44:08 -0700
  • 11bcb53d0a fix mamba models conversion Awni Hannun 2024-10-22 15:32:15 -0700
  • a677638c4b inference works but is hella slow Goekdeniz-Guelmez 2024-10-22 23:06:06 +0200
  • 9ab581d678 notes Goekdeniz-Guelmez 2024-10-22 22:10:53 +0200
  • e43a2ab229 not working, incorrect handling with cache probably Goekdeniz-Guelmez 2024-10-22 22:04:25 +0200
  • 55485b98e8 update Goekdeniz-Guelmez 2024-10-22 21:23:47 +0200
  • d1d480867b
    LoRA: update tools datasets docs (#1063) madroid 2024-10-23 03:19:11 +0800
  • 20015264f4 nits Awni Hannun 2024-10-22 12:18:35 -0700
  • 43d0401bcf nits Awni Hannun 2024-10-22 12:16:59 -0700
  • 758597eaa8 adding multi token input and correct cache handling in ssm step Goekdeniz-Guelmez 2024-10-22 20:44:23 +0200
  • b36625aa6d LoRA: update tools datasets docs madroid 2024-10-23 01:29:45 +0800
  • 66e7bcb886
    override dtype with quant (#1062) Awni Hannun 2024-10-22 09:56:45 -0700
  • 5326d9373a Merge branch 'adding-support-for-mamba2' of https://github.com/Goekdeniz-Guelmez/mlx-examples into adding-support-for-mamba2 Goekdeniz-Guelmez 2024-10-22 18:26:05 +0200
  • b9c57cd429 generation works! trying training now Goekdeniz-Guelmez 2024-10-22 18:25:59 +0200
  • f7bbe458ae Add timeout to generate functions 锦此 2024-10-22 17:06:58 +0800
  • b7078c0f72 override dtype with quant Awni Hannun 2024-10-21 20:12:00 -0700
  • 3263d17ea0 format Awni Hannun 2024-10-21 14:03:15 -0700
  • 645423946a add option to wire model Awni Hannun 2024-10-21 13:35:10 -0700
  • 0ef73f3a2d
    Merge branch 'ml-explore:main' into adding-support-for-mamba2 Gökdeniz Gülmez 2024-10-21 15:14:19 +0200
  • 743763bc2e
    Handle empty string case in maybe_trim_space (#1055) aronson 2024-10-20 22:46:43 -0500
  • d7775e67a6 nit Awni Hannun 2024-10-20 19:20:50 -0700
  • c1634ce81b still generating gibberish Goekdeniz-Guelmez 2024-10-20 18:41:28 +0200
  • ab4cf1d1cf generation works but outputs gibberish Goekdeniz-Guelmez 2024-10-20 18:04:34 +0200
  • 4ab5139c05 quick save Goekdeniz-Guelmez 2024-10-20 16:11:39 +0200
  • d8ca500136
    Handle empty string case in maybe_trim_space Isaac Aronson 2024-10-19 17:46:48 -0500
  • cffd0ed79e run: mlx_lm.server --model mlx-community/Qwen2.5-14B-Instruct-8bit --trust-remote-code --port 8722 chenguangjian.jk 2024-10-20 03:08:27 +0800
  • e6d0a18142 mlx_lm.server --model mlx-community/Qwen2.5-14B-Instruct-8bit --trust-remote-code --port 8722 chenguangjian.jk 2024-10-20 02:18:18 +0800
  • 627b0d8187 mlx_lm.server --model mlx-community/Qwen2.5-14B-Instruct-8bit --trust-remote-code --port 8722 chenguangjian.jk 2024-10-19 22:54:13 +0800
  • 0532ba17fa mlx_lm.server --model mlx-community/Qwen2.5-7B-Instruct-8bit --trust-remote-code --port 8722 chenguangjian.jk 2024-10-19 16:23:49 +0800
  • dbd09720a1 mlx_lm.server --model mlx-community/Ministral-8B-Instruct-2410-8bit --trust-remote-code --port 8722 chenguangjian.jk 2024-10-18 23:33:42 +0800
  • 8a680c3c88
    Merge branch 'ml-explore:main' into main chenguangjian 2024-10-18 23:13:08 +0800
  • dfac3e67fe fix nit in docs Awni Hannun 2024-10-16 12:39:54 -0700
  • cd036ccfb5 fix generation works too (almost) Goekdeniz-Guelmez 2024-10-16 21:13:36 +0200
  • 181d6abedc Merge branch 'adding-support-for-mamba2' of https://github.com/Goekdeniz-Guelmez/mlx-examples into adding-support-for-mamba2 Goekdeniz-Guelmez 2024-10-16 21:09:42 +0200
  • 8073cb486c adding debug statements (somehiw generating only goes through the fist MambaMixer block pass) Goekdeniz-Guelmez 2024-10-16 21:09:30 +0200
  • 855fcc4327
    Merge branch 'ml-explore:main' into adding-support-for-mamba2 Gökdeniz Gülmez 2024-10-16 18:57:55 +0200
  • b899e81589 Merge branch 'main' into flux/config madroid 2024-10-16 21:12:22 +0800
  • f491d473a3
    FLUX: Optimize dataset loading logic (#1038) madroid 2024-10-16 01:37:45 +0800
  • 3d62b058a4
    fix: typo on flux model preloading (#1050) Zak B. Elep 2024-10-16 00:13:01 +0800
  • 93397949d2 fix: typo on flux model preloading Zak B. Elep 2024-10-15 23:55:28 +0800
  • 7df99c6b71 FLUX: save final adapters file madroid 2024-10-15 13:49:44 +0800
  • 130cdae48e FLUX: save train config to json madroid 2024-10-15 13:35:12 +0800
  • 532e961f58 Minor changes Angelos Katharopoulos 2024-10-14 11:39:12 -0700
  • bbd2003047
    FLUX: update README.md (#1036) madroid 2024-10-15 02:21:41 +0800
  • 605c4854f1
    Prompt caching in mlx_lm.server (#1026) Awni Hannun 2024-10-14 10:57:22 -0700
  • 8dca1a2f60
    Tokenizer updates + tests (#1024) Awni Hannun 2024-10-14 10:48:46 -0700
  • 5c4e6ce279 comments Awni Hannun 2024-10-14 10:47:58 -0700
  • 1b05b51dc5 don't throw if no metal Awni Hannun 2024-10-09 19:13:33 -0700
  • d6222ae7ff fix tests Awni Hannun 2024-10-09 13:17:56 -0700
  • d85010bf4b nits Awni Hannun 2024-10-09 12:49:32 -0700
  • cdba586b67 caching in server Awni Hannun 2024-10-09 12:46:44 -0700
  • 6c368f2124
    bump mac tests to use py39 (#1047) Awni Hannun 2024-10-14 10:40:36 -0700
  • c799133998
    Make llm async eval less brittle (#1040) Awni Hannun 2024-10-14 10:25:24 -0700
  • 2dd903b0bf Add push to hub multimodalart 2024-10-14 21:59:56 +0700
  • 3f1c1dde6a
    Merge branch 'ml-explore:main' into adding-support-for-mamba2 Gökdeniz Gülmez 2024-10-14 16:32:00 +0200
  • 64b118852f bump mac tests to use py39 Awni Hannun 2024-10-14 06:39:03 -0700
  • 1e0cda68c6
    Update README.md (#1045) Seitaro Sugawara 2024-10-14 22:21:25 +0900
  • 3d6bd2fe24
    Update musicgen/README.md Awni Hannun 2024-10-14 06:21:01 -0700
  • a6ae63226f
    Update README.md Seitaro Sugawara 2024-10-14 14:06:35 +0900
  • 68518a3194 FLUX: add generate images tqdm desc madroid 2024-10-13 21:47:33 +0800