Add more model support for speculate_decoding and refactor speculate_decoding

[Wanglongzhi2001]

PR types

Others

PR changes

Others

Description

support speculative decoding for Mixtral,Qwen2,Qwen2-moe and refactor speculative decoding.

[paddle-bot]

Thanks for your contribution!

[codecov]

Codecov Report

Attention: Patch coverage is 1.29870% with 76 lines in your changes missing coverage. Please review.

Project coverage is 53.05%. Comparing base (2c1387f) to head (c6deb6f).
Report is 10 commits behind head on develop.

Files with missing lines	Patch %	Lines
...enlp/experimental/transformers/generation_utils.py	0.00%	33 Missing ⚠️
...dlenlp/experimental/transformers/llama/modeling.py	0.00%	10 Missing ⚠️
...enlp/experimental/transformers/mixtral/modeling.py	0.00%	10 Missing ⚠️
...dlenlp/experimental/transformers/qwen2/modeling.py	0.00%	10 Missing ⚠️
...lp/experimental/transformers/qwen2_moe/modeling.py	0.00%	10 Missing ⚠️
paddlenlp/trl/llm_utils.py	0.00%	2 Missing ⚠️
paddlenlp/transformers/auto/modeling.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #9504      +/-   ##
===========================================
+ Coverage    52.85%   53.05%   +0.20%     
===========================================
  Files          700      701       +1     
  Lines       110418   110570     +152     
===========================================
+ Hits         58359    58668     +309     
+ Misses       52059    51902     -157     

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