tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Commit Graph

  • 7037e3c836 deepseekv2 liger support (#1878) Aman Gupta Karmani 2024-08-27 20:52:40 -07:00
  • 467464f045 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-28 03:09:14 +00:00
  • c1a61ae23c fix liger plugin load issues (#1876) Aman Gupta Karmani 2024-08-27 20:08:26 -07:00
  • 86c72380b9 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-28 00:23:33 +00:00
  • 159b8b9a74 monkey-patch transformers to simplify monkey-patching modeling code (#1877) Aman Gupta Karmani 2024-08-27 17:22:26 -07:00
  • 969b16f88f Built site for gh-pages Quarto GHA Workflow Runner 2024-08-27 17:40:18 +00:00
  • 1e43660701 Sample pack trust remote code v2 (#1873) Wing Lian 2024-08-27 13:39:24 -04:00
  • 058d8c1e09 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-27 17:04:04 +00:00
  • f6362d2a05 Add Liger Kernal support for Qwen2 (#1871) Chiwan Park 2024-08-28 02:03:16 +09:00
  • efeaa00bb4 Update docs/dataset-formats/conversation.qmd Chirag Jain 2024-08-27 19:08:54 +05:30
  • 8a84408fc7 Address review comments and add docs Chirag Jain 2024-08-27 04:25:44 +05:30
  • 4805f3ca0a Merge branch 'main' of https://github.com/OpenAccess-AI-Collective/axolotl into cj_tokenizer_default_prompt_template Chirag Jain 2024-08-27 02:35:58 +05:30
  • 68cbda17a0 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-26 19:51:21 +00:00
  • 17af1d7081 clear cuda cache to help with memory leak/creep (#1858) Wing Lian 2024-08-26 15:50:26 -04:00
  • 482fe96a4a Built site for gh-pages Quarto GHA Workflow Runner 2024-08-26 16:57:11 +00:00
  • 2dac1edf72 Fix drop_long_seq bug due to truncation in prompt tokenization strategies when using chat_template (#1867) Chiwan Park 2024-08-27 01:56:12 +09:00
  • 0454b587e1 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-26 16:01:59 +00:00
  • 6819c12cee update specturm authors (#1869) Wing Lian 2024-08-26 12:00:36 -04:00
  • 3e8056b954 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-25 21:54:53 +00:00
  • 8e29bdefdd Spectrum plugin (#1866) Wing Lian 2024-08-25 17:54:02 -04:00
  • 5229c2a73d Built site for gh-pages Quarto GHA Workflow Runner 2024-08-25 16:32:29 +00:00
  • f245964f22 better handling of llama-3 tool rolw (#1782) Wing Lian 2024-08-25 12:31:40 -04:00
  • cb03ff72ab Built site for gh-pages Quarto GHA Workflow Runner 2024-08-24 00:23:59 +00:00
  • 22f4eafa55 simplify logic (#1856) Wing Lian 2024-08-23 20:23:08 -04:00
  • 4f3b5cad43 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-23 21:00:52 +00:00
  • 77a4b9cda2 change up import to prevent AttributeError (#1863) Wing Lian 2024-08-23 17:00:01 -04:00
  • dd0b858185 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-23 18:35:00 +00:00
  • 810ecd4e81 add liger to readme (#1865) Wing Lian 2024-08-23 14:34:03 -04:00
  • f5da86dfa8 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-23 16:38:42 +00:00
  • da0d581a8c add liger example (#1864) Wing Lian 2024-08-23 12:37:50 -04:00
  • 7b598a4ed2 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-23 16:22:49 +00:00
  • 1f686c576c Liger Kernel integration (#1861) Wing Lian 2024-08-23 12:21:51 -04:00
  • 395feebea5 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-23 16:19:50 +00:00
  • e8ff5d5738 don't mess with bnb since it needs compiled wheels (#1859) Wing Lian 2024-08-23 12:18:47 -04:00
  • 9bf03a6d1e Built site for gh-pages Quarto GHA Workflow Runner 2024-08-23 15:41:18 +00:00
  • 328fd4b3b7 add axolotl community license (#1862) Wing Lian 2024-08-23 11:40:21 -04:00
  • 756a34f0fe wip for tp Wing Lian 2024-08-23 10:57:57 -04:00
  • 198f7cd893 2d parallel llama fsdp Wing Lian 2024-08-23 00:02:14 -04:00
  • 5b15816cf4 drop valueerror as this was from when 4bit required gptq remove-gptq-warn Wing Lian 2024-08-22 19:16:32 -04:00
  • 8ee30f5954 Merge branch 'main' into cj_tokenizer_default_prompt_template Chirag Jain 2024-08-23 03:44:25 +05:30
  • f1873957ec Built site for gh-pages Quarto GHA Workflow Runner 2024-08-22 20:40:19 +00:00
  • fefa95e350 most model types now support flash attention 2 regardless of multipack support (#1854) Wing Lian 2024-08-22 16:39:23 -04:00
  • 0b35889407 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-22 17:14:21 +00:00
  • b33dc07a77 rename nightly test and add badge (#1853) Wing Lian 2024-08-22 13:13:33 -04:00
  • a4310969e8 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-22 17:11:47 +00:00
  • dcbff16983 run nightly ci builds against upstream main (#1851) Wing Lian 2024-08-22 13:10:54 -04:00
  • 2f8037fee6 ensure that the hftrainer deepspeed config is set before the trainer class is ever init'ed (#1850) [skip ci] Wing Lian 2024-08-22 13:10:40 -04:00
  • edb70f789c Built site for gh-pages Quarto GHA Workflow Runner 2024-08-22 15:48:28 +00:00
  • de4ea2d1f2 docs: minor syntax highlight fix (#1839) Aman Gupta Karmani 2024-08-22 08:47:34 -07:00
  • 7ed92e61c2 fix: prompt phi (#1845) [skip ci] JohanWork 2024-08-22 17:46:57 +02:00
  • 9caa3eb699 make the train_on_eos default to turn so all eos tokens are treated the same (#1847) [skip ci] Wing Lian 2024-08-22 11:45:37 -04:00
  • 5b0b774e38 ensure that the bias is also in the correct dtype (#1848) [skip ci] Wing Lian 2024-08-22 11:45:00 -04:00
  • c3fc529bfc numpy 2.1.0 was released, but incompatible with numba (#1849) [skip ci] Wing Lian 2024-08-22 11:44:45 -04:00
  • 957c956f89 rename jamba example (#1846) [skip ci] Gal Cohen (galco) 2024-08-22 16:22:55 +03:00
  • d2cf919e4f Built site for gh-pages Quarto GHA Workflow Runner 2024-08-21 17:38:39 +00:00
  • e85d954f52 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-21 17:38:13 +00:00
  • f07802f9fa examples: fix tiny-llama pretrain yml syntax (#1840) Aman Gupta Karmani 2024-08-21 10:37:51 -07:00
  • 8169df172a Built site for gh-pages Quarto GHA Workflow Runner 2024-08-21 17:37:47 +00:00
  • 9f917245f6 feat: add jamba chat_template (#1843) Gal Cohen (galco) 2024-08-21 20:37:17 +03:00
  • 649c19aba3 pretrain: fix with sample_packing=false (#1841) Aman Gupta Karmani 2024-08-21 10:36:51 -07:00
  • e074063ac7 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-20 16:42:40 +00:00
  • 5aac4bc284 fix: dont change quant storage dtype in case of fsdp (#1837) Gal Cohen (galco) 2024-08-20 19:41:48 +03:00
  • bf4795c2a9 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-19 19:00:17 +00:00
  • e29931259b optionally save the final FSDP model as a sharded state dict (#1828) Wing Lian 2024-08-19 14:59:24 -04:00
  • 6ef76f1ace remove custom mistral template Chirag Jain 2024-08-19 15:56:47 +05:30
  • 2e758aed6f Merge branch 'main' into cj_tokenizer_default_prompt_template Chirag Jain 2024-08-19 15:52:04 +05:30
  • 9c922ade89 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-17 01:32:52 +00:00
  • b1d2921222 add validation to prevent 8bit lora finetuning on H100s (#1827) Wing Lian 2024-08-16 21:32:00 -04:00
  • 1f09f48d8f fixed small typo fixtypo sunny 2024-08-16 11:27:56 -04:00
  • b44546df6f fixedtypo sunny 2024-08-16 11:22:50 -04:00
  • 967fbf8152 fixedtypo Sunny 2024-08-16 11:17:52 -04:00
  • c144a1ae65 fixed small typo Sunny 2024-08-16 10:57:42 -04:00
  • 0206b57aa9 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-16 14:42:54 +00:00
  • 803fed3e90 update sklearn versrion, torch compile env vars, don't worry about failure on preprocess load model (#1821) Wing Lian 2024-08-16 10:41:51 -04:00
  • 7b994feb40 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-16 11:52:14 +00:00
  • 68a3c7678a fix: parse model_kwargs (#1825) NanoCode012 2024-08-16 20:51:19 +09:00
  • e834e52dbf Built site for gh-pages Quarto GHA Workflow Runner 2024-08-14 13:47:40 +00:00
  • f18925fb4b fix: parse eager_attention (#1824) NanoCode012 2024-08-14 22:46:46 +09:00
  • 21a2302538 Merge branch 'main' into cj_tokenizer_default_prompt_template Chirag Jain 2024-08-12 10:24:02 +05:30
  • 2b890ead05 fsdp fft loading on meta device fsdp-fft Wing Lian 2024-08-11 22:18:04 -04:00
  • 53791c1c2e Built site for gh-pages Quarto GHA Workflow Runner 2024-08-11 20:28:30 +00:00
  • 1853d6021d bump hf dependencies (#1823) Wing Lian 2024-08-11 16:27:41 -04:00
  • a04e1ec370 Built site for gh-pages Quarto GHA Workflow Runner 2024-08-09 15:51:20 +00:00
  • 0801f239cc fix the incorrect max_length for chat template (#1818) Chiwan Park 2024-08-10 00:50:31 +09:00
  • 54392ac8a6 Attempt to run multigpu in PR CI for now to ensure it works (#1815) [skip ci] Wing Lian 2024-08-09 11:50:13 -04:00
  • 3e2b269d06 update tinyllama to use final instead of checkpoints (#1820) [skip ci] Wing Lian 2024-08-09 10:58:19 -04:00
  • 5ee4b7325f fix z3 leaf configuration when not using lists (#1817) [skip ci] Wing Lian 2024-08-09 10:54:52 -04:00
  • c7b0bee57b Built site for gh-pages Quarto GHA Workflow Runner 2024-08-06 19:26:43 +00:00
  • 70978467a0 skip no commit to main on ci (#1814) Wing Lian 2024-08-06 15:25:54 -04:00
  • 89f382a13a Merge branch 'main' into cj_tokenizer_default_prompt_template Chirag Jain 2024-08-06 21:23:14 +05:30
  • 6674f532ac Built site for gh-pages Quarto GHA Workflow Runner 2024-08-06 14:32:54 +00:00
  • 850f999a76 update peft and transformers (#1811) Wing Lian 2024-08-06 10:32:05 -04:00
  • c56e0a79a5 logging improvements (#1808) [skip ci] Wing Lian 2024-08-06 10:31:50 -04:00
  • 35d5e59d78 set z3 leaf for deepseek v2 (#1809) [skip ci] Wing Lian 2024-08-06 09:30:46 -04:00
  • fbbeb4fee0 remove un-necessary zero-first guard as it's already only called in a parent fn (#1810) [skip ci] Wing Lian 2024-08-06 09:29:23 -04:00
  • fde80ab9ff Built site for gh-pages Quarto GHA Workflow Runner 2024-08-05 17:13:16 +00:00
  • ecdda006de One cycle lr (#1803) Wing Lian 2024-08-05 13:12:05 -04:00
  • b7665c26c8 Update conversation.qmd (#1788) [skip ci] Ben Feuer 2024-08-05 12:44:26 -04:00
  • d60b8e681c Built site for gh-pages Quarto GHA Workflow Runner 2024-08-05 16:43:37 +00:00
  • cb023c70db Update instruct-lora-8b.yml (#1789) [skip ci] Aaditya Ura (looking for PhD Fall’24) 2024-08-05 22:13:20 +05:30