Commit Graph

74 Commits

Author SHA1 Message Date
Adrian Wälchli 45c2fcb341
Add AttributeDict container for Fabric (#18943)
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
Co-authored-by: Jirka Borovec <6035284+Borda@users.noreply.github.com>
2023-11-18 09:25:26 -05:00
Jirka Borovec 6628d8a73a
docs: switch todo to comment (#18893) 2023-11-06 11:50:38 +01:00
thomas chaton f5f4d0a264
Improve map, optimize and StreamingDataset (#18912)
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: thomas <thomas@thomass-MacBook-Pro.local>
2023-11-03 11:28:55 +00:00
Carlos Mocholí ad93f64d35
Bitsandbytes docs improvements (#18903) 2023-11-02 15:50:54 +01:00
Jirka Borovec 1e68c50141
ci: restrict build docs on PR (#18923)
* ci: restrict build docs on PR

* linkcheck_timeout
2023-11-02 09:03:03 -04:00
Carlos Mocholí 800b87eb46
Add throughput utilities to Fabric and the Trainer (#18848) 2023-10-30 17:10:29 +01:00
Adrian Wälchli e02bb391af
Utility to disable all instances of `PossibleUserWarning` (#18744)
Co-authored-by: Jirka Borovec <6035284+Borda@users.noreply.github.com>
2023-10-10 06:53:32 -04:00
Adrian Wälchli 74d402095e
docs: Fix display of navigation tiles in Fabric docs (#18733) 2023-10-06 14:36:01 +02:00
Adrian Wälchli 5d819c91fb
Remove `fsdp_overlap_step_with_backward` in favor of native solution (#18726) 2023-10-06 08:11:41 -04:00
Carlos Mocholí d1f8b0f766
Bitsandbytes docs improvements (#18681)
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
2023-09-30 16:19:11 +02:00
Carlos Mocholí 5120ad20f2
Bitsandbytes precision plugin (#18655)
Co-authored-by: awaelchli <aedu.waelchli@gmail.com>
2023-09-29 19:17:18 +02:00
Jirka Borovec e3a6ac2436
intro granularity for docs build (#18675) 2023-09-29 15:10:45 +02:00
Jirka Borovec 358336268f
enable codespell for docs & fixing +TPU (#18629)
* precommit/codespell

* run

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* disable

* more fixing

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* Apply suggestions from code review

* more fixing

* json

* note

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2023-09-26 11:54:44 -04:00
Adrian Wälchli f83ad093e5
Utility function to check shared filesystem (#18586)
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2023-09-25 15:49:52 -04:00
Jirka Borovec d579cfed57
precommit: unify formatting with prettier (#18605)
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2023-09-25 14:34:41 +02:00
Adrian Wälchli 57f5268eb3
Improve the suggested `num_workers` warning (#18591)
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2023-09-21 09:38:25 -04:00
Adrian Wälchli 8094855137
Avoid passing process group to enable FSDP's hybrid-shard (#18583) 2023-09-19 13:46:24 -04:00
Adrian Wälchli c1ee22a687
Optimize import paths for optional dependencies (#18561)
Co-authored-by: Jirka Borovec <6035284+Borda@users.noreply.github.com>
2023-09-15 17:04:19 -04:00
Jirka Borovec bda974281d
docs: switch Fabric to lai template (#18449)
* docs: switch Fabric to lai template
* imports
2023-09-11 18:43:07 +02:00
Jirka Borovec 565046f337
docs: welcome Fabric like Trainer (#18429)
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: Carlos Mocholí <carlossmocholi@gmail.com>
2023-09-08 10:33:44 -04:00
Carlos Mocholí 756e481969
Support the TransformerEngine precision plugin with the Trainer (#18459)
Co-authored-by: Jirka Borovec <6035284+Borda@users.noreply.github.com>
2023-09-07 19:21:00 +02:00
Adrian Wälchli d3ee410100
Add dedicated docs page for init-module (#18416) 2023-08-28 11:28:38 -04:00
Jirka Borovec 8218097833
docs: updating logos (#18375) 2023-08-28 16:49:28 +02:00
Adrian Wälchli f4825e5778
Extend FSDP guide with checkpointing (#18374) 2023-08-23 20:23:16 +02:00
Jirka Borovec 547e7aa393
docs: 1/3 enable Sphinx nitpicky [fabric] (#18069)
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2023-08-23 10:31:20 +02:00
Jirka Borovec 3f73231d5c
ci/docs: enable not fetch assets (#18333) 2023-08-21 20:22:21 +02:00
Adrian Wälchli 6df43685ee
Revamp model parallel docs (FSDP) (3/n) (#18326) 2023-08-17 15:30:58 -04:00
Adrian Wälchli 03ca31c3d3
Avoid updating the device for XLA FSDP in `Fabric.setup()` [TPU] (#18276) 2023-08-11 22:00:23 -04:00
Carlos Mocholí c83774a109
Update docs about double precision with complex numbers (#18269) 2023-08-10 10:36:55 +02:00
Jirka Borovec 8f29bb561b
bump sphinx to 5.3 (#18204) 2023-08-08 15:32:34 +02:00
Adrian Wälchli 3fd24f9591
Remove outdated warning about loading full-state checkpoints in FSDP (#18208) 2023-08-01 20:06:30 +02:00
Adrian Wälchli 6552d29a12
Document limitations of multi-GPU in Jupyter notebooks (#18132)
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2023-07-24 15:22:16 -04:00
Adrian Wälchli 2af425328a
Add FSDP to the docs glossary (#18121) 2023-07-20 09:07:37 +02:00
Adrian Wälchli 6ab6ab8193
Fabric FSDP documentation guide (#18109) 2023-07-19 18:39:07 +02:00
Carlos Mocholí 071f85842e
Support NVIDIA's Transformer Engine as a precision plugin (#17597) 2023-07-19 18:21:58 +02:00
Adrian Wälchli dab373de54
Support loading a raw PyTorch state-dict checkpoint in Fabric (#18049)
Co-authored-by: Jirka Borovec <6035284+Borda@users.noreply.github.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2023-07-18 14:06:17 -04:00
Jirka Borovec 6b52b84ef8
docs: fallback for restoring right scroll menu (#18108) 2023-07-18 14:06:18 +02:00
Minh-Long Luu (刘明龙) c8656f1a27
docs: `fabric_methods.rst` using deprecated parameters (#18015) 2023-07-07 11:32:09 +02:00
Adrian Wälchli 036bd74e47
Disable video controls in Fabric teaser animations (#17984) 2023-07-04 13:27:54 +02:00
Jirka Borovec 00a0f2e8f4
docs: fetch external sources (#17941)
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2023-07-03 20:16:45 +02:00
Carlos Mocholí 58d2387e0c
Add `Fabric.save(filter=...)` (#17845)
Co-authored-by: Adrian Wälchli <aedu.waelchli@gmail.com>
2023-06-20 18:18:59 +00:00
Adrian Wälchli 1204006637
Document tensor shape requirements in `all_gather` (#17816) 2023-06-15 17:30:02 +02:00
Jirka Borovec 76515140d8
docs: bump sphinx ver. (#17824)
* docs: bump sphinx ver.

* myst-parser

* myst-parser

* myst-parser

* 6.2
2023-06-15 09:37:04 -04:00
Adrian Wälchli 9ff7d7120b
Add `rank_zero_first` utility (#17784)
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2023-06-12 10:32:32 +00:00
Adrian Wälchli 3cfacc7d3e
Document non-strict loading of checkpoints in Fabric (#17765) 2023-06-07 22:24:21 +00:00
Adrian Wälchli 24a3115995
Support empty weight initialization in `Fabric.init_module()` (#17627) 2023-06-07 18:33:53 +00:00
Adrian Wälchli 88cd100369
Address feedback for `Fabric.init_module()` (1/4) (#17721)
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
2023-05-31 14:05:29 +00:00
edenlightning e470997af9
docs: fix links in fabric (#17611) 2023-05-11 22:56:11 +02:00
Carlos Mocholí 54e8095a78
Split `init_module` into `init` + `sharded_model` (#17488) 2023-05-05 15:54:52 +02:00
Carlos Mocholí 6ec9a6bd9e
[TPU] Rename classes to use XLA instead of TPU (#17383) 2023-04-28 12:36:22 -04:00