ONNX 模型库
返回模型

说明文档

<h1 align="center"><strong>更多文本，更少点云：面向3D数据高效的点云-语言理解</strong></h1> <p align="center"> Yuan Tang*  Xu Han*  Xianzhi Li<sup>✝</sup>  Qiao Yu  Jinfeng Xu  Yixue Hao  Long Hu  Min Chen <br> 华中科技大学 华南理工大学 </p> </p>

<p align="center"> <a><strong>AAAI 2025 </strong></a> <a href='https://arxiv.org/pdf/2408.15966'><img src='https://img.shields.io/badge/Paper-Arxiv-red'></a> <a href='https://huggingface.co/YuanTang96/GreenPLM'><img src='https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Model-blue'></a> </p>

<!-- contents with emoji -->

📋 目录

🔍 概述

pic_2

pic_2

  • 我们提出了一项新任务——3D数据高效的点云-语言理解，旨在让大语言模型（LLM）仅需极少的3D数据即可实现鲁棒的3D理解能力。
  • 我们提出了GreenPLM，从全新视角解决这一3D数据受限的任务，通过更多免费文本数据来增强点云-LLM的对齐。
  • 我们引入了600万条T3D数据集，设计了三阶段训练策略，并提出了0M-Pooling模块用于token池化。
  • 我们引入了准确率与3D数据比率（A3DR）来衡量3D数据使用效率，并基于开源LLM建立了评估基准。
  • GreenPLM仅使用12%的3D数据就超越了之前的模型，甚至仅使用文本数据就超越了GPT4Point（使用66万条3D数据），展示了卓越的3D数据效率。

📦 训练与评估

下载项目

本项目的代码、权重和数据集已上传至 Hugging Face。只需下载即可开始使用本项目。

安装环境

进入项目目录并执行以下命令：

conda create -n greenplm python=3.10 -y
conda activate greenplm
bash  envInstall.sh

项目目录介绍

  • ./greenplm/release 包含论文的权重、训练脚本和测试脚本。
  • ./pretrained_weight 存放项目训练和测试阶段所需的预训练权重。
  • ./lava-vicuna_2024_4_Phi-3-mini-4k-instruct 是Phi-3的权重目录。
  • ./dataset/T3D 是本项目提出的600万条数据集。
  • ./dataset/T3D/stage_1/brief_1M_caption.json 是第一阶段的数据集。
  • ./dataset/T3D/stage_2/stage_2_data_210k.json 是第二阶段的数据集。

数据集准备

./dataset/Objaverse/8192_npy.zip 包含本项目所需的Objaverse点云数据。解压数据集：

unzip ./dataset/Objaverse/8192_npy.zip -d ./dataset/Objaverse/

推理

论文权重

GreenPLM-0

仅使用文本数据训练的模型，即（第一阶段 & 第二阶段）。

bash ./release/paper/scripts/test/release_stage_2.sh

输出的JSON结果保存在 ./release/paper/result_json/stage_2。

GreenPLM

使用少量3D数据训练的模型，即（第一阶段 & 第二阶段 & 第三阶段）。

bash ./release/paper/scripts/test/release_stage_3.sh

输出的JSON结果保存在 ./release/paper/result_json/stage_3。

使用完整T3D数据集的权重

<details> <summary>我们还提供了使用完整T3D数据集训练的权重，即在第二阶段使用T3D的500万条数据，而非论文中的21万条。（点击展开）</summary>

GreenPLM-0

仅使用文本数据训练的模型，即（第一阶段 & 第二阶段）。

bash ./release/5M_data_seting/scripts/test/release_5M_stage_2.sh

输出的JSON结果保存在 ./release/5M_data_seting/result_json/stage_2。

GreenPLM

使用少量3D数据训练的模型，即（第一阶段 & 第二阶段 & 第三阶段）。

bash ./release/5M_data_seting/scripts/test/release_5M_stage_3.sh

输出的JSON结果保存在 ./release/5M_data_seting/result_json/stage_3。

</details>

评估

使用LLM

  • 您可以从 阿里云 获取 DASHSCOPE_API_KEY。评估可能需要约9元人民币（约1.3美元）。
  • 如果您有足够的GPU资源，也可以按照 Qwen2 搭建自己的Qwen2-72B-Instruct服务，然后免费评估结果！
  1. 在Objaverse上评估开放词汇分类
export PYTHONPATH=$PWD
export DASHSCOPE_API_KEY=sk-xxx
python ./pointllm/eval/evaluator_opensource_llm_QwenAPI.py  \
        --results_path /path/to/evaluation/PointLLM_brief_description_val_200_GT_Objaverse_classification_prompt0.json  \
        --eval_type open-free-form-classification  \
        --model_type qwen2-72b-instruct \
        --parallel --num_workers 4

export PYTHONPATH=$PWD
export DASHSCOPE_API_KEY=sk-xxx
python ./pointllm/eval/evaluator_opensource_llm_QwenAPI.py  \
        --results_path /path/to/evaluation/PointLLM_brief_description_val_200_GT_Objaverse_classification_prompt1.json  \
        --eval_type open-free-form-classification  \
        --model_type qwen2-72b-instruct \
        --parallel --num_workers 4
  1. 在ModelNet40上评估闭集零样本分类
export PYTHONPATH=$PWD
export DASHSCOPE_API_KEY=sk-xxx
python ./pointllm/eval/evaluator_opensource_llm_QwenAPI.py  \
    --results_path /path/to/evaluation/ModelNet_classification_prompt0.json  \
    --eval_type modelnet-close-set-classification  \
    --model_type qwen2-72b-instruct \
    --parallel --num_workers 4

export PYTHONPATH=$PWD
export DASHSCOPE_API_KEY=sk-xxx
python ./pointllm/eval/evaluator_opensource_llm_QwenAPI.py  \
    --results_path /path/to/evaluation/ModelNet_classification_prompt1.json  \
    --eval_type modelnet-close-set-classification  \
    --model_type qwen2-72b-instruct \
    --parallel --num_workers 4
  1. 在Objaverse上评估物体描述生成
export PYTHONPATH=$PWD
export DASHSCOPE_API_KEY=sk-xxx
python ./pointllm/eval/evaluator_opensource_llm_QwenAPI.py  \
        --results_path /path/to/evaluation/PointLLM_brief_description_val_200_GT_Objaverse_captioning_prompt2.json  \
        --eval_type object-captioning  \
        --model_type qwen2-72b-instruct \
        --parallel --num_workers 4

传统指标评估

对于物体描述生成任务，运行以下命令使用传统指标Sentence-BERT和SimCSE评估模型输出。

CUDA_VISIBLE_DEVICES=0 python pointllm/eval/traditional_evaluator.py --results_path /path/to/evaluation/PointLLM_brief_description_val_200_GT_Objaverse_captioning_prompt2.json

训练

第一阶段

bash ./release/paper/scripts/train/1.sh

第二阶段：GreenPLM-0

bash ./release/paper/scripts/train/2.sh

第三阶段：GreenPLM

bash ./release/paper/scripts/train/3.sh

<details> <summary>我们还提供了使用完整T3D数据集的训练脚本，即在第二阶段使用T3D的500万条数据，而非论文中的21万条。（点击展开）</summary>

第二阶段：GreenPLM-0

bash ./release/5M_data_seting/scripts/train/2.sh

第三阶段：GreenPLM

bash ./release/5M_data_seting/scripts/train/3.sh

</details>

注意：您可以修改脚本中的 --output_dir 参数来设置训练权重的输出目录。

🔗 引用

如果您觉得我们的工作有帮助，请考虑引用：

@inproceedings{tang2025more,
  title={More text, less point: Towards 3d data-efficient point-language understanding},
  author={Tang, Yuan and Han, Xu and Li, Xianzhi and Yu, Qiao and Xu, Jinfeng and Hao, Yixue and Hu, Long and Chen, Min},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={39},
  number={7},
  pages={7284--7292},
  year={2025}
}

📄 许可证

<a rel="license" href="http://creativecommons.org/licenses/by-nc-sa/4.0/"><img alt="Creative Commons License" style="border-width:0" src="https://i.creativecommons.org/l/by-nc-sa/4.0/80x15.png" /></a> <br /> 本作品采用 <a rel="license" href="http://creativecommons.org/licenses/by-nc-sa/4.0/">知识共享署名-非商业性使用-相同方式共享 4.0 国际许可协议</a>。

📚 相关工作

让我们一起推动3D大语言模型的发展！

  • Point-Bind & Point-LLM：将点云与Image-Bind对齐，无需3D指令数据训练即可推理多模态输入。
  • 3D-LLM：使用2D基础模型编码3D点云的多视图图像。
  • PointLLM：将3D点云与LLaVA结合。
  • ShapeLLM：将强大的点云编码器与LLM结合用于具身场景。
  • MiniGPT-3D ：迈出了高效3D-LLM的第一步，仅需单张RTX 3090 GPU和一天的训练时间。

👏 致谢

我们要感谢 PointLLM、Uni3D、Phi-3 和 LLaVA-pp 的作者们出色的工作和代码库。

YuanTang96/GreenPLM

作者 YuanTang96

↓ 9 ♥ 1

创建时间: 2025-05-23 06:28:35+00:00

更新时间: 2025-05-27 11:58:12+00:00

在 Hugging Face 上查看

文件 (372)

.gitattributes
LICENSE
README.md
cog.yaml
config.json
dataset/Objaverse/8192_npy.zip
dataset/Objaverse/PointLLM_brief_description_val_200_GT.json
dataset/Objaverse/PointLLM_complex_50k_brief_40k_all_90k.json
dataset/T3D/stage_1/brief_1M_caption.json
dataset/T3D/stage_2/stage_2_data_210k.json
dataset/T3D/stage_2/stage_2_data_5M.json
dataset/modelnet40_data/modelnet40_test_8192pts_fps.dat
envInstall.sh
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/.gitattributes
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/CODE_OF_CONDUCT.md
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/LICENSE
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/NOTICE.md
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/README.md
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/SECURITY.md
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/added_tokens.json
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/config.json
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/configuration_phi3.py
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/generation_config.json
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/model-00001-of-00002.safetensors
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/model-00002-of-00002.safetensors
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/model.safetensors.index.json
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/modeling_phi3.py
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/sample_finetune.py
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/special_tokens_map.json
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/tokenizer.json
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/tokenizer.model
lava-vicuna_2024_4_Phi-3-mini-4k-instruct/tokenizer_config.json
llava/__init__.py
llava/__pycache__/__init__.cpython-310.pyc
llava/__pycache__/constants.cpython-310.pyc
llava/__pycache__/conversation.cpython-310.pyc
llava/__pycache__/mm_utils.cpython-310.pyc
llava/bpe_simple_vocab_16e6.txt.gz
llava/constants.py
llava/conversation.py
llava/eval/eval_gpt_review.py
llava/eval/eval_gpt_review_bench.py
llava/eval/eval_gpt_review_visual.py
llava/eval/eval_pope.py
llava/eval/eval_science_qa.py
llava/eval/eval_science_qa_gpt4.py
llava/eval/eval_science_qa_gpt4_requery.py
llava/eval/eval_textvqa.py
llava/eval/generate_webpage_data_from_table.py
llava/eval/m4c_evaluator.py
llava/eval/model_qa.py
llava/eval/model_vqa.py
llava/eval/model_vqa_loader.py
llava/eval/model_vqa_mmbench.py
llava/eval/model_vqa_science.py
llava/eval/qa_baseline_gpt35.py
llava/eval/run_llava.py
llava/eval/summarize_gpt_review.py
llava/eval/webpage/figures/alpaca.png
llava/eval/webpage/figures/bard.jpg
llava/eval/webpage/figures/chatgpt.svg
llava/eval/webpage/figures/llama.jpg
llava/eval/webpage/figures/swords_FILL0_wght300_GRAD0_opsz48.svg
llava/eval/webpage/figures/vicuna.jpeg
llava/eval/webpage/index.html
llava/eval/webpage/script.js
llava/eval/webpage/styles.css
llava/mm_utils.py
llava/model/__init__.py
llava/model/__pycache__/__init__.cpython-310.pyc
llava/model/__pycache__/builder.cpython-310.pyc
llava/model/__pycache__/llava_arch.cpython-310.pyc
llava/model/__pycache__/utils.cpython-310.pyc
llava/model/apply_delta.py
llava/model/bpe_simple_vocab_16e6.txt.gz
llava/model/builder.py
llava/model/consolidate.py
llava/model/language_model/__pycache__/llava_phi3.cpython-310.pyc
llava/model/language_model/llava_llama.py
llava/model/language_model/llava_mistral.py
llava/model/language_model/llava_mpt.py
llava/model/language_model/llava_phi3.py
llava/model/llava_arch.py
llava/model/make_delta.py
llava/model/multimodal_encoder/__pycache__/builder.cpython-310.pyc
llava/model/multimodal_encoder/__pycache__/clip_encoder.cpython-310.pyc
llava/model/multimodal_encoder/__pycache__/point_encoder.cpython-310.pyc
llava/model/multimodal_encoder/builder.py
llava/model/multimodal_encoder/clip_encoder.py
llava/model/multimodal_encoder/point_encoder.py
llava/model/multimodal_projector/__pycache__/Qformer.cpython-310.pyc
llava/model/multimodal_projector/__pycache__/builder.cpython-310.pyc
llava/model/multimodal_projector/builder.py
llava/model/utils.py
llava/serve/__init__.py
llava/serve/__pycache__/__init__.cpython-310.pyc
llava/serve/__pycache__/cli.cpython-310.pyc
llava/serve/controller.py
llava/serve/examples/extreme_ironing.jpg
llava/serve/examples/waterview.jpg
llava/serve/gradio_web_server.py
llava/serve/model_worker.py
llava/serve/register_worker.py
llava/serve/sglang_worker.py
llava/serve/test_message.py
llava/train/__pycache__/llava_trainer.cpython-310.pyc
llava/train/__pycache__/train.cpython-310.pyc
llava/train/llama_flash_attn_monkey_patch.py
llava/train/llama_xformers_attn_monkey_patch.py
llava/train/llava_trainer.py
llava/train/train.py
llava/train/train_mem.py
llava/train/train_xformers.py
llava/utils.py
media/3_stage.png
media/T3D.png
pointllm/__init__.py
pointllm/__pycache__/__init__.cpython-310.pyc
pointllm/__pycache__/conversation.cpython-310.pyc
pointllm/__pycache__/utils.cpython-310.pyc
pointllm/conversation.py
pointllm/data/__init__.py
pointllm/data/__pycache__/__init__.cpython-310.pyc
pointllm/data/__pycache__/modelnet.cpython-310.pyc
pointllm/data/__pycache__/object_point_dataset.cpython-310.pyc
pointllm/data/__pycache__/scanobjectNN.cpython-310.pyc
pointllm/data/__pycache__/utils.cpython-310.pyc
pointllm/data/modelnet.py
pointllm/data/modelnet_config/ModelNet40.yaml
pointllm/data/modelnet_config/modelnet40_shape_names_modified.txt
pointllm/data/modelnet_show.py
pointllm/data/object_point_dataset.py
pointllm/data/utils.py
pointllm/data/utils_backup.py
pointllm/eval/eval_modelnet_cls.py
pointllm/eval/eval_objaverse.py
pointllm/eval/evaluator_opensource_llm_QwenAPI.py
pointllm/eval/traditional_evaluator.py
pointllm/eval/utils.py
pointllm/model/__init__.py
pointllm/model/pointllm.py
pointllm/model/utils.py
pointllm/train/llama_flash_attn_monkey_patch.py
pointllm/train/pointllm_trainer.py
pointllm/train/train.py
pointllm/train/train_mem.py
pointllm/utils.py
pointnet++/Pointnet2_PyTorch/.pre-commit-config.yaml
pointnet++/Pointnet2_PyTorch/.travis.yml
pointnet++/Pointnet2_PyTorch/MANIFEST.in
pointnet++/Pointnet2_PyTorch/README.rst
pointnet++/Pointnet2_PyTorch/UNLICENSE
pointnet++/Pointnet2_PyTorch/pointnet2/__init__.py
pointnet++/Pointnet2_PyTorch/pointnet2/_version.py
pointnet++/Pointnet2_PyTorch/pointnet2/config/config.yaml
pointnet++/Pointnet2_PyTorch/pointnet2/config/model/msg.yaml
pointnet++/Pointnet2_PyTorch/pointnet2/config/model/ssg.yaml
pointnet++/Pointnet2_PyTorch/pointnet2/config/task/cls.yaml
pointnet++/Pointnet2_PyTorch/pointnet2/config/task/semseg.yaml
pointnet++/Pointnet2_PyTorch/pointnet2/config/task_model/cls-msg.yaml
pointnet++/Pointnet2_PyTorch/pointnet2/config/task_model/cls-ssg.yaml
pointnet++/Pointnet2_PyTorch/pointnet2/config/task_model/semseg-msg.yaml
pointnet++/Pointnet2_PyTorch/pointnet2/config/task_model/semseg-ssg.yaml
pointnet++/Pointnet2_PyTorch/pointnet2/data/.gitignore
pointnet++/Pointnet2_PyTorch/pointnet2/data/Indoor3DSemSegLoader.py
pointnet++/Pointnet2_PyTorch/pointnet2/data/ModelNet40Loader.py
pointnet++/Pointnet2_PyTorch/pointnet2/data/__init__.py
pointnet++/Pointnet2_PyTorch/pointnet2/data/data_utils.py
pointnet++/Pointnet2_PyTorch/pointnet2/models/__init__.py
pointnet++/Pointnet2_PyTorch/pointnet2/models/pointnet2_msg_cls.py
pointnet++/Pointnet2_PyTorch/pointnet2/models/pointnet2_msg_sem.py
pointnet++/Pointnet2_PyTorch/pointnet2/models/pointnet2_ssg_cls.py
pointnet++/Pointnet2_PyTorch/pointnet2/models/pointnet2_ssg_sem.py
pointnet++/Pointnet2_PyTorch/pointnet2/train.py
pointnet++/Pointnet2_PyTorch/pointnet2/utils/.gitignore
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/MANIFEST.in
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/__init__.py
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/include/ball_query.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/include/cuda_utils.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/include/group_points.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/include/interpolate.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/include/sampling.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/include/utils.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/ball_query.cpp
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/ball_query_gpu.cu
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/bindings.cpp
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/group_points.cpp
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/group_points_gpu.cu
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/interpolate.cpp
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/interpolate_gpu.cu
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/sampling.cpp
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/sampling_gpu.cu
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_ext.cpython-310-x86_64-linux-gnu.so
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/_version.py
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/pointnet2_modules.py
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/lib.linux-x86_64-cpython-310/pointnet2_ops/pointnet2_utils.py
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/.ninja_deps
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/.ninja_log
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/build.ninja
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/ball_query.o
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/ball_query_gpu.o
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/bindings.o
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/group_points.o
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/group_points_gpu.o
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/interpolate.o
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/interpolate_gpu.o
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/sampling.o
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/build/temp.linux-x86_64-cpython-310/pointnet2_ops/_ext-src/src/sampling_gpu.o
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/dist/pointnet2_ops-3.0.0-py3.10-linux-x86_64.egg
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops.egg-info/PKG-INFO
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops.egg-info/SOURCES.txt
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops.egg-info/dependency_links.txt
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops.egg-info/requires.txt
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops.egg-info/top_level.txt
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/__init__.py
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/include/ball_query.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/include/cuda_utils.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/include/group_points.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/include/interpolate.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/include/sampling.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/include/utils.h
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/ball_query.cpp
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/ball_query_gpu.cu
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/bindings.cpp
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/group_points.cpp
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/group_points_gpu.cu
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/interpolate.cpp
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/interpolate_gpu.cu
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/sampling.cpp
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_ext-src/src/sampling_gpu.cu
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/_version.py
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/pointnet2_modules.py
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/pointnet2_ops/pointnet2_utils.py
pointnet++/Pointnet2_PyTorch/pointnet2_ops_lib/setup.py
pointnet++/Pointnet2_PyTorch/pyproject.toml
pointnet++/Pointnet2_PyTorch/requirements.txt
pointnet++/Pointnet2_PyTorch/setup.py
pointnet++/Pointnet2_PyTorch/tests/conftest.py
pointnet++/Pointnet2_PyTorch/tests/test_cls.py
pointnet++/Pointnet2_PyTorch/tests/test_semseg.py
pointnet++/Pointnet2_PyTorch/tox.ini
pretrained_weight/Uni3D_PC_encoder/.gitattributes
pretrained_weight/Uni3D_PC_encoder/modelzoo/uni3d-base/model.pt
pretrained_weight/Uni3D_PC_encoder/modelzoo/uni3d-giant/model.pt
pretrained_weight/Uni3D_PC_encoder/modelzoo/uni3d-large/model.pt
pretrained_weight/Uni3D_PC_encoder/modelzoo/uni3d-small/model.pt
pretrained_weight/Uni3D_PC_encoder/modelzoo/uni3d-tiny/model.pt
pretrained_weight/clip_used_in_Uni3D/.gitattributes
pretrained_weight/clip_used_in_Uni3D/README.md
pretrained_weight/clip_used_in_Uni3D/merges.txt
pretrained_weight/clip_used_in_Uni3D/open_clip_config.json
pretrained_weight/clip_used_in_Uni3D/open_clip_pytorch_model.bin
pretrained_weight/clip_used_in_Uni3D/special_tokens_map.json
pretrained_weight/clip_used_in_Uni3D/tokenizer.json
pretrained_weight/clip_used_in_Uni3D/tokenizer_config.json
pretrained_weight/clip_used_in_Uni3D/vocab.json
pretrained_weight/eval_model_weight/all-mpnet-base-v2/.gitattributes
pretrained_weight/eval_model_weight/all-mpnet-base-v2/1_Pooling/config.json
pretrained_weight/eval_model_weight/all-mpnet-base-v2/README.md
pretrained_weight/eval_model_weight/all-mpnet-base-v2/config.json
pretrained_weight/eval_model_weight/all-mpnet-base-v2/config_sentence_transformers.json
pretrained_weight/eval_model_weight/all-mpnet-base-v2/data_config.json
pretrained_weight/eval_model_weight/all-mpnet-base-v2/model.safetensors
pretrained_weight/eval_model_weight/all-mpnet-base-v2/modules.json
pretrained_weight/eval_model_weight/all-mpnet-base-v2/onnx/model.onnx ONNX
pretrained_weight/eval_model_weight/all-mpnet-base-v2/onnx/model_O1.onnx ONNX
pretrained_weight/eval_model_weight/all-mpnet-base-v2/onnx/model_O2.onnx ONNX
pretrained_weight/eval_model_weight/all-mpnet-base-v2/onnx/model_O3.onnx ONNX
pretrained_weight/eval_model_weight/all-mpnet-base-v2/onnx/model_O4.onnx ONNX
pretrained_weight/eval_model_weight/all-mpnet-base-v2/onnx/model_qint8_arm64.onnx ONNX
pretrained_weight/eval_model_weight/all-mpnet-base-v2/onnx/model_qint8_avx512.onnx ONNX
pretrained_weight/eval_model_weight/all-mpnet-base-v2/onnx/model_qint8_avx512_vnni.onnx ONNX
pretrained_weight/eval_model_weight/all-mpnet-base-v2/onnx/model_quint8_avx2.onnx ONNX
pretrained_weight/eval_model_weight/all-mpnet-base-v2/openvino/openvino_model.bin
pretrained_weight/eval_model_weight/all-mpnet-base-v2/openvino/openvino_model.xml
pretrained_weight/eval_model_weight/all-mpnet-base-v2/openvino/openvino_model_qint8_quantized.bin
pretrained_weight/eval_model_weight/all-mpnet-base-v2/openvino/openvino_model_qint8_quantized.xml
pretrained_weight/eval_model_weight/all-mpnet-base-v2/pytorch_model.bin
pretrained_weight/eval_model_weight/all-mpnet-base-v2/sentence_bert_config.json
pretrained_weight/eval_model_weight/all-mpnet-base-v2/special_tokens_map.json
pretrained_weight/eval_model_weight/all-mpnet-base-v2/tokenizer.json
pretrained_weight/eval_model_weight/all-mpnet-base-v2/tokenizer_config.json
pretrained_weight/eval_model_weight/all-mpnet-base-v2/train_script.py
pretrained_weight/eval_model_weight/all-mpnet-base-v2/vocab.txt
pretrained_weight/eval_model_weight/sup-simcse-roberta-large/.gitattributes
pretrained_weight/eval_model_weight/sup-simcse-roberta-large/README.md
pretrained_weight/eval_model_weight/sup-simcse-roberta-large/config.json
pretrained_weight/eval_model_weight/sup-simcse-roberta-large/flax_model.msgpack
pretrained_weight/eval_model_weight/sup-simcse-roberta-large/merges.txt
pretrained_weight/eval_model_weight/sup-simcse-roberta-large/pytorch_model.bin
pretrained_weight/eval_model_weight/sup-simcse-roberta-large/special_tokens_map.json
pretrained_weight/eval_model_weight/sup-simcse-roberta-large/tokenizer_config.json
pretrained_weight/eval_model_weight/sup-simcse-roberta-large/vocab.json
pyproject.toml
release/5M_data_seting/scripts/test/release_5M_stage_2.sh
release/5M_data_seting/scripts/test/release_5M_stage_3_lr_1e5.sh
release/5M_data_seting/scripts/train/2.sh
release/5M_data_seting/scripts/train/3.sh
release/5M_data_seting/weight/stage_2/5M_low_lr_1e5/README.md
release/5M_data_seting/weight/stage_2/5M_low_lr_1e5/adapter_config.json
release/5M_data_seting/weight/stage_2/5M_low_lr_1e5/adapter_model.safetensors
release/5M_data_seting/weight/stage_2/5M_low_lr_1e5/config.json
release/5M_data_seting/weight/stage_2/5M_low_lr_1e5/non_lora_trainables.bin
release/5M_data_seting/weight/stage_2/5M_low_lr_1e5/trainer_state.json
release/5M_data_seting/weight/stage_3/use_stage_5M_data_lr_1e5/README.md
release/5M_data_seting/weight/stage_3/use_stage_5M_data_lr_1e5/adapter_config.json
release/5M_data_seting/weight/stage_3/use_stage_5M_data_lr_1e5/adapter_model.safetensors
release/5M_data_seting/weight/stage_3/use_stage_5M_data_lr_1e5/config.json
release/5M_data_seting/weight/stage_3/use_stage_5M_data_lr_1e5/non_lora_trainables.bin
release/5M_data_seting/weight/stage_3/use_stage_5M_data_lr_1e5/trainer_state.json
release/paper/scripts/test/release_stage_2.sh
release/paper/scripts/test/release_stage_3.sh
release/paper/scripts/train/1.sh
release/paper/scripts/train/2.sh
release/paper/scripts/train/3.sh
release/paper/weight/stage_1/config.json
release/paper/weight/stage_1/mm_projector.bin
release/paper/weight/stage_1/trainer_state.json
release/paper/weight/stage_2/README.md
release/paper/weight/stage_2/adapter_config.json
release/paper/weight/stage_2/adapter_model.safetensors
release/paper/weight/stage_2/config.json
release/paper/weight/stage_2/non_lora_trainables.bin
release/paper/weight/stage_2/trainer_state.json
release/paper/weight/stage_3/README.md
release/paper/weight/stage_3/adapter_config.json
release/paper/weight/stage_3/adapter_model.safetensors
release/paper/weight/stage_3/config.json
release/paper/weight/stage_3/non_lora_trainables.bin
release/paper/weight/stage_3/trainer_state.json
scripts/convert_gqa_for_eval.py
scripts/convert_mmbench_for_submission.py
scripts/convert_mmvet_for_eval.py
scripts/convert_seed_for_submission.py
scripts/convert_sqa_to_llava.py
scripts/convert_sqa_to_llava_base_prompt.py
scripts/convert_vizwiz_for_submission.py
scripts/convert_vqav2_for_submission.py
scripts/extract_mm_projector.py
scripts/finetune.sh
scripts/finetune_full_schedule.sh
scripts/finetune_lora.sh
scripts/finetune_qlora.sh
scripts/finetune_sqa.sh
scripts/merge_lora_weights.py
scripts/pretrain.sh
scripts/pretrain_xformers.sh
scripts/sqa_eval_batch.sh
scripts/sqa_eval_gather.sh
scripts/upload_pypi.sh
scripts/v1_5/eval/gqa.sh
scripts/v1_5/eval/llavabench.sh
scripts/v1_5/eval/mmbench.sh
scripts/v1_5/eval/mmbench_cn.sh
scripts/v1_5/eval/mme.sh
scripts/v1_5/eval/mmvet.sh
scripts/v1_5/eval/pope.sh
scripts/v1_5/eval/qbench.sh
scripts/v1_5/eval/qbench_zh.sh
scripts/v1_5/eval/seed.sh
scripts/v1_5/eval/sqa.sh
scripts/v1_5/eval/textvqa.sh
scripts/v1_5/eval/vizwiz.sh
scripts/v1_5/eval/vqav2.sh
scripts/v1_5/finetune.sh
scripts/v1_5/finetune_lora.sh
scripts/v1_5/finetune_task.sh
scripts/v1_5/finetune_task_lora.sh
scripts/v1_5/pretrain.sh
scripts/zero2.json
scripts/zero3.json
scripts/zero3_offload.json