├── .gitignore ├── LiteSys ├── README.md ├── README.pdf ├── assets │ ├── hf_sglang.png │ ├── litesys.png │ ├── qwen_trial.png │ └── qwen_trials.png ├── examples │ ├── dist_engine_generate.py │ ├── engine_generate.py │ ├── eval_lcb.py │ ├── eval_lcb.sh │ ├── eval_lcb_dist.py │ ├── eval_lcb_dist.sh │ ├── eval_math.py │ ├── eval_math.sh │ ├── eval_math_dist.py │ ├── eval_math_dist.sh │ ├── example_dp.sh │ ├── example_tp.sh │ ├── grade_lcb.py │ ├── ref_hf.py │ └── ref_hf.sh ├── litesys │ ├── __init__.py │ ├── attention │ │ ├── __init__.py │ │ ├── batch_cache.py │ │ ├── cache_utils.py │ │ └── sparsity_utils.py │ ├── engine │ │ ├── __init__.py │ │ ├── dist_engine.py │ │ ├── engine.py │ │ └── utils.py │ ├── helper.py │ ├── logging_config.py │ ├── models │ │ ├── __init__.py │ │ ├── auto_model.py │ │ ├── base.py │ │ ├── model_utils.py │ │ ├── qwen.py │ │ ├── qwen3.py │ │ ├── qwen3_dist.py │ │ ├── qwen3_moe_dist.py │ │ ├── qwen_dist.py │ │ └── qwen_layer.py │ └── utils.py └── requirements.txt ├── README.md ├── assets ├── AIME24-topk-genlen.jpg ├── AIME24-topk-trial.jpg ├── Bolt.png ├── F1A.png └── F1B.png ├── benchmark ├── README.md ├── blocktopk.py ├── blocktopk_decoder.py ├── dense.py ├── dense_decoder.py └── model_utils.py ├── cost_model ├── best_of_N │ ├── compare_sparse_numTrials.ipynb │ ├── cost_model_ntrial.py │ └── frontier_numTrials.ipynb ├── long_CoT │ ├── compare_sparse_genlen.ipynb │ ├── cost_model_genlen.py │ └── frontier_genlen.ipynb └── utils.py ├── install.sh └── requirements.txt /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/.gitignore -------------------------------------------------------------------------------- /LiteSys/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/README.md -------------------------------------------------------------------------------- /LiteSys/README.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/README.pdf -------------------------------------------------------------------------------- /LiteSys/assets/hf_sglang.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/assets/hf_sglang.png -------------------------------------------------------------------------------- /LiteSys/assets/litesys.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/assets/litesys.png -------------------------------------------------------------------------------- /LiteSys/assets/qwen_trial.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/assets/qwen_trial.png -------------------------------------------------------------------------------- /LiteSys/assets/qwen_trials.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/assets/qwen_trials.png -------------------------------------------------------------------------------- /LiteSys/examples/dist_engine_generate.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/dist_engine_generate.py -------------------------------------------------------------------------------- /LiteSys/examples/engine_generate.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/engine_generate.py -------------------------------------------------------------------------------- /LiteSys/examples/eval_lcb.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/eval_lcb.py -------------------------------------------------------------------------------- /LiteSys/examples/eval_lcb.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/eval_lcb.sh -------------------------------------------------------------------------------- /LiteSys/examples/eval_lcb_dist.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/eval_lcb_dist.py -------------------------------------------------------------------------------- /LiteSys/examples/eval_lcb_dist.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/eval_lcb_dist.sh -------------------------------------------------------------------------------- /LiteSys/examples/eval_math.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/eval_math.py -------------------------------------------------------------------------------- /LiteSys/examples/eval_math.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/eval_math.sh -------------------------------------------------------------------------------- /LiteSys/examples/eval_math_dist.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/eval_math_dist.py -------------------------------------------------------------------------------- /LiteSys/examples/eval_math_dist.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/eval_math_dist.sh -------------------------------------------------------------------------------- /LiteSys/examples/example_dp.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/example_dp.sh -------------------------------------------------------------------------------- /LiteSys/examples/example_tp.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/example_tp.sh -------------------------------------------------------------------------------- /LiteSys/examples/grade_lcb.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/grade_lcb.py -------------------------------------------------------------------------------- /LiteSys/examples/ref_hf.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/ref_hf.py -------------------------------------------------------------------------------- /LiteSys/examples/ref_hf.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/examples/ref_hf.sh -------------------------------------------------------------------------------- /LiteSys/litesys/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /LiteSys/litesys/attention/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /LiteSys/litesys/attention/batch_cache.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/attention/batch_cache.py -------------------------------------------------------------------------------- /LiteSys/litesys/attention/cache_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/attention/cache_utils.py -------------------------------------------------------------------------------- /LiteSys/litesys/attention/sparsity_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/attention/sparsity_utils.py -------------------------------------------------------------------------------- /LiteSys/litesys/engine/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /LiteSys/litesys/engine/dist_engine.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/engine/dist_engine.py -------------------------------------------------------------------------------- /LiteSys/litesys/engine/engine.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/engine/engine.py -------------------------------------------------------------------------------- /LiteSys/litesys/engine/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/engine/utils.py -------------------------------------------------------------------------------- /LiteSys/litesys/helper.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/helper.py -------------------------------------------------------------------------------- /LiteSys/litesys/logging_config.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/logging_config.py -------------------------------------------------------------------------------- /LiteSys/litesys/models/__init__.py: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /LiteSys/litesys/models/auto_model.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/models/auto_model.py -------------------------------------------------------------------------------- /LiteSys/litesys/models/base.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/models/base.py -------------------------------------------------------------------------------- /LiteSys/litesys/models/model_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/models/model_utils.py -------------------------------------------------------------------------------- /LiteSys/litesys/models/qwen.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/models/qwen.py -------------------------------------------------------------------------------- /LiteSys/litesys/models/qwen3.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/models/qwen3.py -------------------------------------------------------------------------------- /LiteSys/litesys/models/qwen3_dist.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/models/qwen3_dist.py -------------------------------------------------------------------------------- /LiteSys/litesys/models/qwen3_moe_dist.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/models/qwen3_moe_dist.py -------------------------------------------------------------------------------- /LiteSys/litesys/models/qwen_dist.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/models/qwen_dist.py -------------------------------------------------------------------------------- /LiteSys/litesys/models/qwen_layer.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/models/qwen_layer.py -------------------------------------------------------------------------------- /LiteSys/litesys/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/litesys/utils.py -------------------------------------------------------------------------------- /LiteSys/requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/LiteSys/requirements.txt -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/README.md -------------------------------------------------------------------------------- /assets/AIME24-topk-genlen.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/assets/AIME24-topk-genlen.jpg -------------------------------------------------------------------------------- /assets/AIME24-topk-trial.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/assets/AIME24-topk-trial.jpg -------------------------------------------------------------------------------- /assets/Bolt.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/assets/Bolt.png -------------------------------------------------------------------------------- /assets/F1A.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/assets/F1A.png -------------------------------------------------------------------------------- /assets/F1B.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/assets/F1B.png -------------------------------------------------------------------------------- /benchmark/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/benchmark/README.md -------------------------------------------------------------------------------- /benchmark/blocktopk.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/benchmark/blocktopk.py -------------------------------------------------------------------------------- /benchmark/blocktopk_decoder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/benchmark/blocktopk_decoder.py -------------------------------------------------------------------------------- /benchmark/dense.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/benchmark/dense.py -------------------------------------------------------------------------------- /benchmark/dense_decoder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/benchmark/dense_decoder.py -------------------------------------------------------------------------------- /benchmark/model_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/benchmark/model_utils.py -------------------------------------------------------------------------------- /cost_model/best_of_N/compare_sparse_numTrials.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/cost_model/best_of_N/compare_sparse_numTrials.ipynb -------------------------------------------------------------------------------- /cost_model/best_of_N/cost_model_ntrial.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/cost_model/best_of_N/cost_model_ntrial.py -------------------------------------------------------------------------------- /cost_model/best_of_N/frontier_numTrials.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/cost_model/best_of_N/frontier_numTrials.ipynb -------------------------------------------------------------------------------- /cost_model/long_CoT/compare_sparse_genlen.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/cost_model/long_CoT/compare_sparse_genlen.ipynb -------------------------------------------------------------------------------- /cost_model/long_CoT/cost_model_genlen.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/cost_model/long_CoT/cost_model_genlen.py -------------------------------------------------------------------------------- /cost_model/long_CoT/frontier_genlen.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/cost_model/long_CoT/frontier_genlen.ipynb -------------------------------------------------------------------------------- /cost_model/utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/cost_model/utils.py -------------------------------------------------------------------------------- /install.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/install.sh -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Infini-AI-Lab/Kinetics/HEAD/requirements.txt --------------------------------------------------------------------------------