sarasa 0.0.2__py3-none-any.whl

sarasa/utils.py

@@ -0,0 +1,163 @@
+import contextlib
+import gc
+import os
+import sys
+import time
+from datetime import timedelta
+from functools import cache
+
+import torch
+from loguru import logger
+from torch import distributed as dist
+from torch import nn
+
+
+def setup_logger(config) -> None:
+    logger.remove()
+    if config.debug:
+        logger_format = f"<blue>RANK={rank()}</blue> | " + (
+            "<green>{time:YYYY-MM-DD HH:mm:ss.SSS}</green> | "
+            "<level>{level: <8}</level> | "
+            "<cyan>{name}</cyan>:<cyan>{function}</cyan>:<cyan>{line}</cyan> | "
+            "<level>{message}</level>"
+        )
+        logger.add(
+            sys.stderr,
+            format=logger_format,
+            backtrace=True,
+            diagnose=True,
+            level="DEBUG",
+        )
+    else:
+        # log to stderr only for rank 0
+        logger.add(sys.stderr, backtrace=True, diagnose=True, level="INFO" if rank() == 0 else "ERROR")
+
+
+@contextlib.contextmanager
+def set_dtype(
+    dtype: torch.dtype,
+) -> None:
+    old_dtype = torch.get_default_dtype()
+    torch.set_default_dtype(dtype)
+    try:
+        yield
+    finally:
+        torch.set_default_dtype(old_dtype)
+
+
+class GarbageCollector:
+    def __init__(
+        self,
+        gc_freq: int,
+    ) -> None:
+        self.gc_freq = gc_freq
+        if self.gc_freq > 0:
+            # manually manage gc
+            gc.disable()
+
+    def collect(
+        self,
+        step: int,
+    ) -> None:
+        if self.gc_freq <= 0:
+            # auto gc, nothing to do
+            return
+
+        if step % self.gc_freq == 0:
+            begin = time.perf_counter()
+            gc.collect(generation=1)
+            end = time.perf_counter()
+            logger.info(f"Garbage collection at step {step} took {end - begin:.4f} seconds")
+
+
+# distributed utils
+
+
+@cache
+def world_size() -> int:
+    if "WORLD_SIZE" in os.environ:
+        return int(os.environ["WORLD_SIZE"])
+
+    if not dist.is_initialized():
+        return 1
+    return dist.get_world_size()
+
+
+@cache
+def rank() -> int:
+    if "RANK" in os.environ:
+        return int(os.environ["RANK"])
+
+    if not dist.is_initialized():
+        return 0
+    return dist.get_rank()
+
+
+@cache
+def local_rank() -> int:
+    if not dist.is_initialized():
+        return 0
+    return dist.get_local_rank()
+
+
+def init_distributed(
+    backend: str,
+    init_timeout_seconds: int,
+) -> None:
+    if "RANK" in os.environ:
+        # run with torchrun
+        dist.init_process_group(
+            backend=backend,
+            timeout=timedelta(seconds=init_timeout_seconds),
+        )
+        logger.info("Initialized distributed process group")
+    else:
+        logger.info("Skipping distributed initialization")
+
+
+def update_timeout(
+    timeout_seconds: int,
+    device: torch.device,
+) -> None:
+    logger.info(f"Updating distributed timeout to {timeout_seconds} seconds")
+    torch.distributed.barrier(device_ids=[torch.accelerator.current_device_index()])
+    torch.accelerator.synchronize(device)
+
+    # at the moment, default process group is the only one supported (None)
+    torch.distributed.distributed_c10d._set_pg_timeout(timedelta(seconds=timeout_seconds), None)
+
+
+def apply_distributed(
+    config,
+    model: nn.Module,
+    device: torch.device,
+    compile: bool,
+) -> None:
+    mesh = dist.device_mesh.init_device_mesh(device.type, (world_size(),))
+
+    if config.name == "ddp":
+        from torch.distributed._composable.replicate import replicate
+
+        if compile:
+            torch._dynamo.config.optimize_ddp = "ddp_optimizer"
+
+        replicate(model, device_mesh=mesh, bucket_cap_mb=100)
+        logger.info("Applied DDP to the model")
+
+    elif config.name == "fsdp":
+        from torch.distributed.fsdp import MixedPrecisionPolicy, fully_shard
+
+        # todo: make dtypes configurable
+        mp_policy = MixedPrecisionPolicy(
+            param_dtype=torch.bfloat16,
+            reduce_dtype=torch.float32,
+        )
+
+        for block in model.blocks:
+            fully_shard(block, mesh=mesh, mp_policy=mp_policy, reshard_after_forward=config.reshard_after_forward)
+        fully_shard(model, mesh=mesh, mp_policy=mp_policy, reshard_after_forward=config.reshard_after_forward)
+
+        logger.info(
+            f"Applied FSDP to the model (param_dtype={mp_policy.param_dtype}, "
+            f"reduce_dtype={mp_policy.reduce_dtype}, reshard_after_forward={config.reshard_after_forward})"
+        )

sarasa-0.0.2.dist-info/METADATA

@@ -0,0 +1,138 @@
+Metadata-Version: 2.4
+Name: sarasa
+Version: 0.0.2
+Summary: Add your description here
+License-File: LICENSE
+Requires-Python: >=3.13
+Requires-Dist: datasets>=4.5.0
+Requires-Dist: loguru>=0.7.3
+Requires-Dist: numpy>=2.4.1
+Requires-Dist: rich>=14.2.0
+Requires-Dist: tensorboard>=2.20.0
+Requires-Dist: tokenizers>=0.22.2
+Requires-Dist: tyro>=1.0.5
+Provides-Extra: cpu
+Requires-Dist: torch>=2.10.0; extra == 'cpu'
+Provides-Extra: cu128
+Requires-Dist: torch>=2.10.0; extra == 'cu128'
+Provides-Extra: cu130
+Requires-Dist: torch>=2.10.0; extra == 'cu130'
+Provides-Extra: flash-attn
+Requires-Dist: flash-attn-cute; extra == 'flash-attn'
+Description-Content-Type: text/markdown
+
+# sarasa
+
+A minimum LLM training framework built on pure PyTorch with simplicity and extensibility.
+
+## Installation
+
+```bash
+uv sync [--extra cpu|cu128|cu130] [--extra flash_attn]
+```
+
+or
+
+```bash
+uv add sarasa[cpu|cu128|cu130]
+```
+
+## Features
+
+- Pure PyTorch implementation
+- Flexible configuration system with command-line overrides
+- Support from a single GPU to multiple GPUs (simple DDP and FSDP for now)
+- Selective activation checkpointing (SAC) for memory efficiency
+- Async distributed checkpoint saving
+
+- [ ] Checkpoint loading
+
+## Usage
+
+It's (almost) ready to use.
+First, set up tokenizer, e.g.,
+
+```bash
+mkdir tokenizer
+cd tokenizer
+uvx hf download --local-dir . --include "tokenizer*" "meta-llama/Llama-3.1-8B"
+```
+
+Then, the following command starts training of a GPT model on FineWeb-edu with a single or multiple GPUs.
+
+```bash
+uv run torchrun --nproc_per_node="gpu" main.py \
+--config-file configs/example.py \
+[--train.local-batch-size 8 ...] # override config options as needed
+```
+
+### Extending with Custom Components
+
+Extending Sarasa is as simple as defining your own configuration dataclasses with `create` methods for custom models, optimizers, data loaders, etc. 
+Here's an example of using a custom optimizer:
+
+```python
+from sarasa import Trainer, Config
+
+class CustomOptimizer(torch.optim.Optimizer):
+    ...
+
+class CustomOptim:
+    lr: float = ...
+
+    def create(self,
+               model: torch.nn.Module
+    ) -> torch.optim.Optimizer:
+        return CustomOptimizer(model.parameters(), lr=self.lr, ...)
+
+class CustomOptim2:
+    lr: float = ...
+
+    def create(self,
+               model: torch.nn.Module
+    ) -> torch.optim.Optimizer:
+        return CustomOptimizer(model.parameters(), lr=self.lr, ...)
+
+
+if __name__ == "__main__":
+    config = Config.from_cli(optim_type=CustomOptim | CustomOptim2)
+    trainer = Trainer(config)
+    trainer.train()
+```
+
+From the command line, you can specify which custom optimizer to use:
+
+```bash
+python script.py optim:custom_optim --optim.lr 0.001 ...
+```
+
+### Config File Example
+
+It's very simple. IDE autocompletion will help you.
+
+```python
+from sarasa.config import Config, Data, LRScheduler, Model, Train, LRScheduler
+from custom_optim import CustomOptim
+
+# only one Config instance should be defined in each config file
+config = Config.create(
+    model=Model(num_layers=12),
+    train=Train(
+        local_batch_size=16,
+        global_batch_size=256,
+        dtype="bfloat16",
+    ),
+    optim=CustomOptim(lr=0.001),
+    lr_scheduler=LRScheduler(
+        decay_type="linear",
+        warmup_steps=1000,
+        total_steps=100000,
+    ),
+    data=Data(tokenizer_path="./tokenizer"),
+    seed=12,
+)
+```
+
+## Acknowledgements
+
+This project is heavily inspired by and borrows code from `torchtitan`.

sarasa-0.0.2.dist-info/RECORD

@@ -0,0 +1,21 @@
+sarasa/__init__.py,sha256=PrMUKyqcTKgKR3R3VMTzfuQ9EwMNVyw7qRl84Dz3d28,77
+sarasa/activation_checkpoint.py,sha256=iGib3e2GFxBLOtgcPLQZnzw0Ru6Gd_yFqWZmUw0Cfa4,3056
+sarasa/checkpoint.py,sha256=nZNo-qv3hvtzZuN0xw4SsCP4QmIM7E5nqGBxGfGYZo0,3616
+sarasa/config.py,sha256=7flyQPS_Ati0xk9qP4opUJTn1wxssZpLh4DbplUFP8k,8570
+sarasa/metrics.py,sha256=OzTuK3Oed-I_2FC7rrE9FYi3NgTdKsDsVkWlGgJGh0M,10636
+sarasa/trainer.py,sha256=0wK2QRZuwtmOA-gxD-nmAUY-sEA0C0aAmSjlufxP7HE,9212
+sarasa/utils.py,sha256=iPteVmAWmEzvriAeUL36d3MKvp27m2oJO2yFNOO_ufk,4398
+sarasa/data/__init__.py,sha256=I0JOb9QrHEj9zXUX8kLir6ONAyiozeagzApig0WcSt8,1150
+sarasa/data/hf_datasets.py,sha256=DUlCpBOcDtZNEGrx4AtZTPW5IMtxIXMX_pKfnQEqzEg,3966
+sarasa/data/tokenizer.py,sha256=JhUOl9USJRM-DVPY02ouiaNUhAu1w2LLGquMnAyyA68,1752
+sarasa/models/__init__.py,sha256=w9p4lZ0oEH2kRMxPh88Ogphwqx_o_Ik8Upfv9SrW7hA,3223
+sarasa/models/attention.py,sha256=rWm6NurkS5wdnzP_LPonCMJt_gQulySQDPFpVZtpGWU,3006
+sarasa/models/llama3.py,sha256=jGrrC2AQJvdyo_YvbGC4vmy23k-19Itsj_fSHUY2QTc,4509
+sarasa/models/nanochat_gpt.py,sha256=cpyoXwlWhqeUtgYSQ673AW6lh9BQ-64_9G9XURfZ1MY,8721
+sarasa/models/utils.py,sha256=_0F8yFVB2ZVClr8YppBhPOWtpNF0dSkmaElCYCkO_co,1111
+sarasa/optimizers/__init__.py,sha256=TH7CV-dexzVIm_NJKpo3VxnzwLvUjPyckD_oXMT48xo,1844
+sarasa/optimizers/utils.py,sha256=yI1_yHllJFyGbFW8jdMbvLfa5k7zUXjdgkvbr64mFOI,705
+sarasa-0.0.2.dist-info/METADATA,sha256=oLIgc94HZ54wJTLmb1f6BhcTL4wy8TFtz-CCW0G7ji8,3452
+sarasa-0.0.2.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+sarasa-0.0.2.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+sarasa-0.0.2.dist-info/RECORD,,

sarasa-0.0.2.dist-info/WHEEL

@@ -0,0 +1,4 @@
+Wheel-Version: 1.0
+Generator: hatchling 1.28.0
+Root-Is-Purelib: true
+Tag: py3-none-any

sarasa-0.0.2.dist-info/licenses/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.