├── .gitignore ├── LICENSE ├── README.md ├── assets └── fig1.png ├── configs └── config.yaml ├── decoding_tree_sketching ├── __init__.py ├── kvbatch_decoder.py ├── run_experiments.py └── utils │ ├── __init__.py │ └── eval_utils.py ├── inference_example.py ├── notebooks ├── README.md └── example_DeepSeek_R1_Distill_Qwen_1_5B.ipynb ├── pyproject.toml ├── result └── fig │ ├── deepseek-qwen3-1.5B-acc.png │ ├── deepseek-qwen3-1.5B-repetition.png │ ├── deepseek-qwen3-7B-acc.png │ └── deepseek-qwen3-7B-repetition.png └── scripts ├── run_all_dts.sh └── run_all_std.sh /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/.gitignore -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/LICENSE -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/README.md -------------------------------------------------------------------------------- /assets/fig1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/assets/fig1.png -------------------------------------------------------------------------------- /configs/config.yaml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/configs/config.yaml -------------------------------------------------------------------------------- /decoding_tree_sketching/__init__.py: -------------------------------------------------------------------------------- 1 | 2 | -------------------------------------------------------------------------------- /decoding_tree_sketching/kvbatch_decoder.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/decoding_tree_sketching/kvbatch_decoder.py -------------------------------------------------------------------------------- /decoding_tree_sketching/run_experiments.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/decoding_tree_sketching/run_experiments.py -------------------------------------------------------------------------------- /decoding_tree_sketching/utils/__init__.py: -------------------------------------------------------------------------------- 1 | 2 | -------------------------------------------------------------------------------- /decoding_tree_sketching/utils/eval_utils.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/decoding_tree_sketching/utils/eval_utils.py -------------------------------------------------------------------------------- /inference_example.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/inference_example.py -------------------------------------------------------------------------------- /notebooks/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/notebooks/README.md -------------------------------------------------------------------------------- /notebooks/example_DeepSeek_R1_Distill_Qwen_1_5B.ipynb: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/notebooks/example_DeepSeek_R1_Distill_Qwen_1_5B.ipynb -------------------------------------------------------------------------------- /pyproject.toml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/pyproject.toml -------------------------------------------------------------------------------- /result/fig/deepseek-qwen3-1.5B-acc.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/result/fig/deepseek-qwen3-1.5B-acc.png -------------------------------------------------------------------------------- /result/fig/deepseek-qwen3-1.5B-repetition.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/result/fig/deepseek-qwen3-1.5B-repetition.png -------------------------------------------------------------------------------- /result/fig/deepseek-qwen3-7B-acc.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/result/fig/deepseek-qwen3-7B-acc.png -------------------------------------------------------------------------------- /result/fig/deepseek-qwen3-7B-repetition.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/result/fig/deepseek-qwen3-7B-repetition.png -------------------------------------------------------------------------------- /scripts/run_all_dts.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/scripts/run_all_dts.sh -------------------------------------------------------------------------------- /scripts/run_all_std.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ZichengXu/Decoding-Tree-Sketching/HEAD/scripts/run_all_std.sh --------------------------------------------------------------------------------