12 Pull requests merged by 9 people

• Add twenty redundant data in post pretrain

  #8777 merged Jul 18, 2024

• [BUG] Add ten redundant data in post pretrain

  #8776 merged Jul 18, 2024

• fix incorrect token counting in llm/predictor.py

  #8769 merged Jul 18, 2024

• Enable parallel_config to use commas as delimiters.

  #8677 merged Jul 17, 2024

• Add dataloader_drop_last to eval dataloader

  #8773 merged Jul 17, 2024

• fix the ce for the unittest

  #8772 merged Jul 17, 2024

• Add the missing truncation=True in llm/predictor.py

  #8768 merged Jul 17, 2024

• add benchmark baichuan2 scripts

  #8683 merged Jul 17, 2024

• 【inference】support load or save Llama2-7b in three patterns

  #8766 merged Jul 16, 2024

• [DEV] Update develop version show

  #8754 merged Jul 15, 2024

• [NPU] Fix sequence parallel lib import

  #8760 merged Jul 15, 2024

• [AutoParallel] update loss for global clip

  #8750 merged Jul 12, 2024

9 Pull requests opened by 7 people

• [Fix] Enable tensor parallel tests.

  #8757 opened Jul 12, 2024

• Refine savable

  #8758 opened Jul 12, 2024

• support Llama2 fp8c16 inference

  #8762 opened Jul 14, 2024

• Hetertrain

  #8763 opened Jul 15, 2024

• [AutoParallel] Refine save load

  #8767 opened Jul 16, 2024

• [Model] Add resized_lm_head

  #8770 opened Jul 16, 2024

• [Prediction] Update LLM prediction.

  #8778 opened Jul 18, 2024

• [CodeStyle] remove markdownlint-cli

  #8779 opened Jul 18, 2024

• sz-ernie4-1 [moe] fix save under not `use-hybrid-model`

  #8782 opened Jul 18, 2024

7 Issues closed by 4 people

• [Bug]: 昇腾npu训练uie-base模型报错

  #8355 closed Jul 17, 2024

• 修改paddlenlp logger的等级

  #6081 closed Jul 16, 2024

• [Feature]: Implement `Meta-Transformer`: A Unified Framework for Multimodal Learning

  #7182 closed Jul 14, 2024

• [Question]: uie-x 循环推理显存增长问题

  #7560 closed Jul 14, 2024

• fluid operator conditional_block_infer 问题

  #8275 closed Jul 13, 2024

• [Bug]: RUN python export_model.py --model_name_or_path meta-llama/Llama-2-7b --output_path ./llama2-7b-static --dtype float16 --inference_model

  #7495 closed Jul 13, 2024

• [Bug]: PaddleNLP 3.0 Llama2 模型 显存占用变多

  #8756 closed Jul 12, 2024

8 Issues opened by 8 people

• [Bug]: 例子（信息抽取）报错

  #8780 opened Jul 18, 2024

• [Question]: 模型下载路径？

  #8775 opened Jul 18, 2024

• 在V100上的WINT8量化

  #8774 opened Jul 18, 2024

• [Question]: 示例代码出错

  #8771 opened Jul 17, 2024

• [Question]: 调用paddlenlp的uie模型，报错Segmentation fault (core dumped)

  #8764 opened Jul 16, 2024

• [Question]: bleu.py出现ZeroDivisionError

  #8761 opened Jul 14, 2024

• [Question]: 是否支持java接口调用模型推理，并集成到web服务中？

  #8759 opened Jul 12, 2024

• [Question]: rocketqa模型

  #8755 opened Jul 12, 2024

24 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• 如何对长文本进行抽取

  #8434 commented on Jul 13, 2024 • 0 new comments

• [Bug]: get_rank_by_dim_and_process_id 函数未实现

  #8428 commented on Jul 13, 2024 • 0 new comments

• [Bug]: TaskFlow zero_shot_text_classification

  #8427 commented on Jul 13, 2024 • 0 new comments

• [Question]:在xpu上使用taskflow预测时间过长

  #8349 commented on Jul 13, 2024 • 0 new comments

• [Question]: gpt3的lora训练问题

  #8081 commented on Jul 13, 2024 • 0 new comments

• [Question]: 给予一段话，判断该文章是否提及过类似这段话的内容？文章很长，而话很短，考虑到用Transformer好像不太靠谱。请问有什么方法可以建议一下？

  #8080 commented on Jul 13, 2024 • 0 new comments

• [Question]: 在使用2.7.2版本跑llama2-7b的lora遇到关于dataset的问题（报错：An error occurred while generating the dataset）

  #8069 commented on Jul 13, 2024 • 0 new comments

• [Question]: UIE信息抽取做 关系抽取任务，出现幻想，文本中没有出现的信息，会出现幻觉

  #8068 commented on Jul 13, 2024 • 0 new comments

• [Question]: paddle.distributed.launch 启动多进程训练结束后Loading best model from checkpoint 报错

  #8429 commented on Jul 15, 2024 • 0 new comments

• [Question]: 再使用UIE-X封闭域信息抽取时遇到的问题

  #8293 commented on Jul 15, 2024 • 0 new comments

• [Question]: 求助，chatglm2 单卡sft内存溢出

  #8612 commented on Jul 15, 2024 • 0 new comments

• [Question]: Data annotation and pre processing for Relation Extraction

  #8457 commented on Jul 16, 2024 • 0 new comments

• [Question]: 分布式

  #8453 commented on Jul 16, 2024 • 0 new comments

• [Bug]: ImportError: DLL load failed while importing libpaddle: 找不到指定的程序。

  #8449 commented on Jul 16, 2024 • 0 new comments

• [Bug]: pipelines的document_stores.faiss中vector_id更新出错

  #8346 commented on Jul 16, 2024 • 0 new comments

• tp推理，内存溢出问题 #8656

  #8736 commented on Jul 16, 2024 • 0 new comments

• [Question]: 在V100上实现WINT8量化性能降低

  #8738 commented on Jul 17, 2024 • 0 new comments

• 文档信息抽取uie-x-base推理速度和模型加载问题咨询

  #8684 commented on Jul 17, 2024 • 0 new comments

• [NVIDIA] TE Integration

  #7229 commented on Jul 15, 2024 • 0 new comments

• vocabulary_expansion_pretraining

  #7755 commented on Jul 18, 2024 • 0 new comments

• 解决HuggingFace 预训练模型压缩部署示例中自动压缩x2paddle_sst2问题

  #8118 commented on Jul 14, 2024 • 0 new comments

• FP8 PTQ With Physical Dependency 0515

  #8443 commented on Jul 16, 2024 • 0 new comments

• [CI] Move tests/llm to unit test gpu.

  #8599 commented on Jul 12, 2024 • 0 new comments

• Fix final train loss.

  #8718 commented on Jul 12, 2024 • 0 new comments