[LLM] Update model convert and fix TP for deepseekv3

[DrownFish19]

Before submitting

• Lint code. If there are lint issues, please format the code first.

# Install and register `pre-commit` in the project folder
pip install pre-commit && pre-commit install

# Process previous code files separately
pre-commit run --file XXXX.py

• Add test cases into tests folder. If there are codecov issues, please add tests cases first.

PR types

Bug fixes

PR changes

Models

Description

1. 为_get_name_mapping增加e_score_correction_bias。
2. 转换最后一层（MTP（Eagle）参数）以用于推理
   • DeepseekV3 包含 embed_tokens.weight 和 layers.61.embed_tokens.weight的共享权重。
   • PaddleNLP 从 tp_action 中为embed_tokens.weight 和 layers.61.embed_tokens.weight随机选择action，这会导致错误选择并且遗漏其中一个。
   • 修改匹配逻辑，将tp_action对应key按照从长到短排序，优先匹配长key，保证匹配正确。
3. 修复 DeepseekV2MLP 中的 MoE 模块
4. 新增 Deepseek-R1系列参数

[codecov]

Codecov Report

Attention: Patch coverage is 4.16667% with 23 lines in your changes missing coverage. Please review.

Project coverage is 52.21%. Comparing base (fb3e4c0) to head (a0f0535).
Report is 10 commits behind head on develop.

Files with missing lines	Patch %	Lines
paddlenlp/transformers/deepseek_v2/modeling.py	4.54%	21 Missing ⚠️
paddlenlp/transformers/conversion_utils.py	0.00%	2 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #9797      +/-   ##
===========================================
+ Coverage    51.64%   52.21%   +0.57%     
===========================================
  Files          741      730      -11     
  Lines       119141   115812    -3329     
===========================================
- Hits         61525    60477    -1048     
+ Misses       57616    55335    -2281     

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

[yuanlehome]

这一块建议yong用num_nextn_predict_layers来决定逻辑，未来v4可能不一定只有1层～

[DrownFish19]

已经修改

[ZHUI]

LGTM