[FUCK] remove fastsafetensors

[yuanlehome]

Motivation

💡 If this PR is a Cherry Pick, the PR title needs to follow the format by adding the [Cherry-Pick] label at the very beginning and appending the original PR ID at the end. For example, [Cherry-Pick][CI] Add check trigger and logic(#5191)

💡 如若此PR是Cherry Pick，PR标题需遵循格式，在最开始加上[Cherry-Pick]标签，以及最后面加上原PR ID，例如[Cherry-Pick][CI] Add check trigger and logic(#5191)

*** for #5330

Modifications

Usage or Command

Accuracy Tests

Checklist

• Add at least a tag in the PR title.
  • Tag list: [[FDConfig],[APIServer],[Engine], [Scheduler], [PD Disaggregation], [Executor], [Graph Optimization], [Speculative Decoding], [RL], [Models], [Quantization], [Loader], [OP], [KVCache], [DataProcessor], [BugFix], [Docs], [CI], [Optimization], [Feature], [Benchmark], [Others], [XPU], [HPU], [GCU], [DCU], [Iluvatar], [Metax]]
  • You can add new tags based on the PR content, but the semantics must be clear.
• Format your code, run pre-commit before commit.
• Add unit tests. Please write the reason in this PR if no unit tests.
• Provide accuracy results.
• If the current PR is submitting to the release branch, make sure the PR has been submitted to the develop branch, then cherry-pick it to the release branch with the [Cherry-Pick] PR tag.

[Copilot]

Pull request overview

This pull request removes the fastsafetensors dependency from the codebase and consolidates on using the standard safetensors library (version >=0.7.0). The PR eliminates the custom fastsafetensors loading mechanism and associated workarounds, simplifying the weight loading infrastructure.

Key Changes:

• Removed fastsafetensors dependency from all requirements files and updated safetensors to version >=0.7.0
• Removed fastsafetensors_weights_iterator and load_tp_checkpoint_v1 functions from weight loading utilities
• Removed use_fastsafetensor configuration option and related environment variable
• Removed the _patch_fastsafetensors() workaround function from package initialization

Reviewed changes

Copilot reviewed 9 out of 10 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
requirements_metaxgpu.txt	Removed fastsafetensors==0.1.14, updated safetensors from ==0.7.0rc0 to >=0.7.0
requirements_iluvatar.txt	Removed fastsafetensors==0.1.14, updated safetensors from ==0.7.0rc0 to >=0.7.0
requirements_dcu.txt	Removed fastsafetensors==0.1.14, updated safetensors from ==0.7.0rc0 to >=0.7.0
requirements.txt	Updated safetensors from >=0.7.0rc0 to >=0.7.0
fastdeploy/model_executor/load_weight_utils.py	Removed fastsafetensors imports and related functions; simplified load_composite_checkpoint to use standard safetensors loading only
fastdeploy/envs.py	Removed FD_USE_FASTSAFETENSOR environment variable
fastdeploy/engine/engine.py	Updated regex pattern to remove "fastsafetensors" from checkpoint loading progress detection
fastdeploy/engine/async_llm.py	Updated regex pattern to remove "fastsafetensors" from checkpoint loading progress detection
fastdeploy/config.py	Removed use_fastsafetensor configuration field from LoadConfig
fastdeploy/init.py	Removed subprocess and sys imports; removed _patch_fastsafetensors() function and its invocation

---

Critical Issues with PR Metadata:

1. PR Title: The title "[FUCK] remove fastsafetensors" contains unprofessional language. According to the project's PR guidelines, titles should follow the format [TAG]Title where TAG should be one of the standard tags like [BugFix], [Feature], [Optimization], etc. Please change the title to something professional like [Optimization] Remove fastsafetensors dependency or [Feature] Remove fastsafetensors dependency.

2. PR Description: The PR description is incomplete. According to the guidelines, it should include:

   • Motivation: Why is fastsafetensors being removed? (e.g., "The fastsafetensors library is no longer needed as safetensors>=0.7.0 provides all required functionality")
   • Modifications: A clear list of what was changed (e.g., removed fastsafetensors dependency, removed related code, updated safetensors version constraints)
   • Testing: How this change was tested to ensure no regression

Please update the PR metadata before merging.

[Copilot]

The function signature for load_pre_sharded_checkpoint has been modified to remove the use_fastsafetensor parameter. However, the docstring still just says "load_pre_sharded_checkpoint" without describing the function's purpose or its parameters. Consider updating the docstring to reflect the current function signature and provide a meaningful description of what the function does and what parameters it expects.

Suggested change

	load_pre_sharded_checkpoint
	Loads a pre-sharded checkpoint for a given model and local rank.
	Args:
	model_path (str): Path to the model directory containing checkpoint shards.
	local_rank (int): The local rank (shard index) to load.
	Returns:
	dict: A state dictionary mapping parameter names to their corresponding tensors.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 33.33333% with 2 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (develop@f88c159). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
fastdeploy/model_executor/load_weight_utils.py	33.33%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             develop    #5371   +/-   ##
==========================================
  Coverage           ?   59.78%           
==========================================
  Files              ?      325           
  Lines              ?    40207           
  Branches           ?     6086           
==========================================
  Hits               ?    24039           
  Misses             ?    14276           
  Partials           ?     1892           

Flag	Coverage Δ
GPU	59.78% <33.33%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[paddle-bot]

Thanks for your contribution!