Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / ymcui/Chinese-LLaMA-Alpaca-2 issues and pull requests
#575 - lm_datasets = lm_datasets.train_test_split(test_size = data_args.validation_split_percentage)
Issue -
State: open - Opened by dshwei 14 days ago
- 1 comment
#574 - llama.cpp更新后的wiki使用
Issue -
State: closed - Opened by Jianliang-Shen about 2 months ago
- 2 comments
Labels: stale
#573 - 为什么llama的回答特别地乱
Issue -
State: closed - Opened by 327635328 2 months ago
- 3 comments
Labels: stale
#572 - chinese-llama-2-13b-hf可否直接用bf16继续预训练?
Issue -
State: closed - Opened by NLP-Learning 2 months ago
- 4 comments
Labels: stale
#397 - Update requirements.txt
Pull Request -
State: open - Opened by reterVision 10 months ago
- 3 comments
#114 - 预训练及指令微调对于结构化数据是怎么处理的
Issue -
State: closed - Opened by zhangjiawei5911 about 1 year ago
- 2 comments
Labels: stale
#112 - Langchain的最后一项环境准备下载不了pip install faiss-gpu==1.7.2
Issue -
State: closed - Opened by YLlllllllllll about 1 year ago
- 3 comments
#111 - 使用手动转化,合并lora模型报错
Issue -
State: closed - Opened by noending about 1 year ago
- 2 comments
Labels: stale
#110 - inference with lora 时报错expected scalar type Half but found Float
Issue -
State: closed - Opened by arceus-jia about 1 year ago
- 5 comments
Labels: stale
#109 - 请问一下,Chinese-LLaMA-2-7B模型只包含lora权重,没有重新训练embed_tokens,lm_head吗?
Issue -
State: closed - Opened by litchiyj about 1 year ago
- 2 comments
Labels: stale
#108 - 推理准确问题
Issue -
State: closed - Opened by goog about 1 year ago
- 4 comments
Labels: stale
#107 - 指令精调报错
Issue -
State: closed - Opened by cccgw about 1 year ago
- 7 comments
#106 - 是否可以自己训练llama2_70b?
Issue -
State: closed - Opened by Lyn4ever29 about 1 year ago
- 7 comments
Labels: stale
#105 - 请问基于transformer的gradio_demo有输入示例吗,自行测试似乎不对。
Issue -
State: closed - Opened by xiaoqi25478 about 1 year ago
- 3 comments
Labels: stale
#103 - Add privateGPT support
Pull Request -
State: closed - Opened by ymcui about 1 year ago
#102 - Chinese-Alpaca-2-LoRA-7B是怎么训练得到的?
Issue -
State: closed - Opened by weilx2267 about 1 year ago
- 2 comments
#101 - add new tokenizer merge recipe
Issue -
State: closed - Opened by enpassanty about 1 year ago
- 3 comments
Labels: stale
#100 - [FR]能否提供和一代一样的privateGPT融合部署方式?
Issue -
State: closed - Opened by 1-2-3 about 1 year ago
- 2 comments
#99 - 多机多卡训练出现:torch.cuda.OutOfMemoryError: CUDA out of memory.
Issue -
State: closed - Opened by Double-bear about 1 year ago
- 5 comments
Labels: stale
#98 - Update README_vllm.md
Pull Request -
State: closed - Opened by GoGoJoestar about 1 year ago
#97 - Exception: Current loss scale already at minimum - cannot decrease scale anymore. Exiting run.
Issue -
State: closed - Opened by YinSonglin1997 about 1 year ago
- 2 comments
Labels: stale
#96 - 指令精调训练报错
Issue -
State: closed - Opened by ai499 about 1 year ago
- 18 comments
Labels: stale
#95 - qlora使用fp16训练出现loss nan
Issue -
State: closed - Opened by yanghh2000 about 1 year ago
- 4 comments
#94 - llama.app项目中调用模型转换失败
Issue -
State: closed - Opened by SolarTorch about 1 year ago
- 11 comments
Labels: stale
#92 - 指令精调启动报错
Issue -
State: closed - Opened by PaulHuang01 about 1 year ago
- 10 comments
#91 - Add CFG sampling
Pull Request -
State: closed - Opened by airaria about 1 year ago
- 1 comment
#90 - langchain_qa.py啥也不返回
Issue -
State: closed - Opened by xxm1668 about 1 year ago
- 10 comments
Labels: stale
#89 - 使用langchain加载模型失败
Issue -
State: closed - Opened by zicheqingluo about 1 year ago
- 6 comments
Labels: stale
#87 - 能否也给一下合并后的sha256 值,我这推理有问题,不知是否是合并的问题
Issue -
State: closed - Opened by icowan about 1 year ago
- 5 comments
Labels: stale
#86 - RuntimeError: Failed to import transformers.models.llama.modeling_llama because of the following error (look up to see its traceback): /usr/local/lib/python3.10/dist-packages/transformer_engine_extensions.cpython-310-x86_64-linux-gnu.so: undefined symbol: _ZN3c10ltERKNS_6SymIntEi
Issue -
State: closed - Opened by gpww about 1 year ago
- 2 comments
Labels: stale
#84 - How to get PT/SFT training dataset?
Issue -
State: closed - Opened by athenawisdoms about 1 year ago
- 2 comments
Labels: stale
#82 - 模型预训练用的llama-alpaca2基座训练的,PT用的txt文本1.5G的文本,只训练lora参数,为何最后得到的 adapter_model.bin 只有1K 为啥。
Issue -
State: closed - Opened by musellama about 1 year ago
- 6 comments
Labels: stale
#81 - 無法利用LangChain.CTransformers 載入Chinese-Llama-2-7b ggml 模型
Issue -
State: closed - Opened by wennycooper about 1 year ago
- 4 comments
#80 - Improve the error message of the merging script
Pull Request -
State: closed - Opened by airaria about 1 year ago
#79 - Add support for LangChain
Pull Request -
State: closed - Opened by iMountTai about 1 year ago
#78 - expected scalar type Half but found Float
Issue -
State: closed - Opened by Faysir about 1 year ago
- 2 comments
#77 - 合併 LoRa model 不成功, 沒有產生最終模型檔
Issue -
State: closed - Opened by wennycooper about 1 year ago
- 2 comments
#76 - 开启 gradient checkpointing 和 flash-attn2 时 lora sft 在 eval 时报错:"use_cache is not supported"
Issue -
State: closed - Opened by tarnish233 about 1 year ago
- 9 comments
Labels: stale
#75 - Add --verbose argument to the C-Eval script
Pull Request -
State: closed - Opened by airaria about 1 year ago
#74 - Add quantization results on Chinese-LLaMA-2-7B
Pull Request -
State: closed - Opened by ymcui about 1 year ago
#72 - 只对lora进行精调最后合并报了错
Issue -
State: closed - Opened by icowan about 1 year ago
- 3 comments
#71 - 请问是否考虑训练一个 extended context 版本的模型?
Issue -
State: closed - Opened by jamesljl about 1 year ago
- 4 comments
Labels: stale
#70 - 预训练使用flashattn报错:RuntimeError: shape '[1, 1024, 64, 128]' is invalid for input of size 1048576
Issue -
State: closed - Opened by Double-bear about 1 year ago
- 8 comments
Labels: stale
#69 - Update default decoding hyperparameters
Pull Request -
State: closed - Opened by airaria about 1 year ago
#68 - sha256值对应不上
Issue -
State: closed - Opened by icoderzqliu about 1 year ago
- 1 comment
#67 - 预训练chinese-llama-2-7b时出错
Issue -
State: closed - Opened by desu9 about 1 year ago
- 6 comments
Labels: stale
#66 - 多卡微调时报错:ERROR:torch.distributed.elastic.multiprocessing.api:failed (exitcode: -7) local_rank: 0 (pid: 1280) of binary: /root/miniconda3/envs/llama2/bin/python
Issue -
State: closed - Opened by thugbobby about 1 year ago
- 6 comments
Labels: stale
#65 - add support for text-generation-webui
Pull Request -
State: closed - Opened by iMountTai about 1 year ago
#64 - chinese-alpaca-2训练数据格式问题
Issue -
State: closed - Opened by icoderzqliu about 1 year ago
- 1 comment
#63 - tokenizer是不是没对应好?
Issue -
State: closed - Opened by Zombiessss about 1 year ago
- 6 comments
Labels: stale
#62 - 模型推理的速度很慢
Issue -
State: closed - Opened by amwork2020 about 1 year ago
- 6 comments
Labels: stale
#61 - Fixed llama pre-training bug
Pull Request -
State: closed - Opened by iMountTai about 1 year ago
- 2 comments
#60 - Add system prompt input to Gradio demo
Pull Request -
State: closed - Opened by airaria about 1 year ago
#59 - Add server examples for llama.cpp
Pull Request -
State: closed - Opened by ymcui about 1 year ago
#58 - deepspeed是哪个版本
Issue -
State: closed - Opened by chensongcan about 1 year ago
- 3 comments
Labels: stale
#57 - 请问llama2-7b的显存要求是多少
Issue -
State: closed - Opened by AlexasXu about 1 year ago
- 2 comments
#56 - 预训练完毕,合成完毕,在oobabooga运行模型会出现自问自答的情况。
Issue -
State: closed - Opened by musellama about 1 year ago
- 18 comments
Labels: stale
#55 - Add a screencast of Alpaca-2-7B-q6_k
Pull Request -
State: closed - Opened by ymcui about 1 year ago
#54 - 微调验证损失趋势
Issue -
State: closed - Opened by Daniel-1997 about 1 year ago
- 2 comments
Labels: stale
#53 - 预训练完毕后,进行lora合成报错,只训练lora参数,但是用你们提供的alpaca-lora 合成是OK的
Issue -
State: closed - Opened by musellama about 1 year ago
- 11 comments
Labels: stale
#52 - 感觉中文答非所问,你们的会有这个问题吗?
Issue -
State: closed - Opened by tmpuserx about 1 year ago
- 3 comments
#51 - Add Chinese LLaMA-2/Alpaca-2 tokenizer
Pull Request -
State: closed - Opened by airaria about 1 year ago
#50 - 预训练完美运行,但是保存lora模型为1k
Issue -
State: closed - Opened by msn199959 about 1 year ago
- 1 comment
#48 - fix bug in gradio_demo.py when launch vLLM server
Pull Request -
State: closed - Opened by GoGoJoestar about 1 year ago
#47 - add support for 4bit inference
Pull Request -
State: closed - Opened by iMountTai about 1 year ago
- 3 comments
#46 - Add Colab-based Gradio web demo
Pull Request -
State: closed - Opened by ymcui about 1 year ago
#45 - 扩充词汇表是怎样操作的?
Issue -
State: closed - Opened by jyC23333 about 1 year ago
- 4 comments
Labels: stale
#44 - 请问增量训练数据量大概用了多少B token数呢?
Issue -
State: closed - Opened by peiyingxin about 1 year ago
- 2 comments
Labels: stale
#43 - Streaming openai api support
Pull Request -
State: closed - Opened by yunhaoli24 about 1 year ago
- 17 comments
#42 - Fix system prompt and improve messages
Pull Request -
State: closed - Opened by airaria about 1 year ago
#40 - 关于flashattention增加后的精度分析
Issue -
State: closed - Opened by shiqingzhangCSU about 1 year ago
- 2 comments
Labels: stale
#39 - 关于基于llama版本底座预训练模型,全量sft loss炸裂的问题
Issue -
State: closed - Opened by lucasjinreal about 1 year ago
- 1 comment
#38 - add FlashAttention-2 support
Pull Request -
State: closed - Opened by iMountTai about 1 year ago
- 1 comment
#37 - chinese-alpace-2-7b推理时,为什么要先输出一遍问题再回答?
Issue -
State: closed - Opened by sun1092469590 about 1 year ago
- 15 comments
Labels: stale
#35 - add vLLM surpport for gradio demo, inference script and openai api demo
Pull Request -
State: closed - Opened by GoGoJoestar about 1 year ago
- 2 comments
#34 - 你好,请问这里说的指令精调和RLHF事一个东西吗
Issue -
State: closed - Opened by QJShan about 1 year ago
- 4 comments
Labels: stale
#33 - Add Alpaca-2-7B output examples
Pull Request -
State: closed - Opened by ymcui about 1 year ago
#32 - add arguments for setting system prompts
Pull Request -
State: closed - Opened by airaria about 1 year ago
- 1 comment
#31 - Add a new system prompt that produces longer responses
Pull Request -
State: closed - Opened by ymcui about 1 year ago
#30 - 新扩充的词表模型会开源么?
Issue -
State: closed - Opened by jiejie1993 about 1 year ago
- 8 comments
Labels: stale
#29 - 请问llama2扩充词表的脚本和Chinese-LLaMa-Alpaca项目里面的扩充词表方式是否一模一样呢?
Issue -
State: closed - Opened by thelongestusernameofall about 1 year ago
- 3 comments
#28 - 对比Baichuan7B
Issue -
State: closed - Opened by lucasjinreal about 1 year ago
- 12 comments
#27 - 进行预训练的时候报错。
Issue -
State: closed - Opened by musellama about 1 year ago
- 20 comments
Labels: stale
#26 - 请问如何在该项目基础上做微调时启用FlashAttention-2的高效注意力?
Issue -
State: closed - Opened by RethinkFun about 1 year ago
- 3 comments
Labels: stale
#25 - 后续会开源13B的中文模型吗
Issue -
State: closed - Opened by sun1092469590 about 1 year ago
- 11 comments
Labels: stale
#22 - 请问120G中文语料包括什么内容?增量训练添加英文语料是否更合适呢?
Issue -
State: closed - Opened by peiyingxin about 1 year ago
- 11 comments
Labels: stale
#21 - 请问如果想做全量微调的话,和Lora微调的代码一样吗?
Issue -
State: closed - Opened by changyuying about 1 year ago
- 6 comments
Labels: stale
#19 - 请问下 “ 基于FlashAttention-2的高效注意力” 如何实现的?我在training代码里面没有找到
Issue -
State: closed - Opened by tanguofu about 1 year ago
- 5 comments
#17 - 请问继续pretrain和sft都是按照4k长度来进行的吗?谢谢!
Issue -
State: closed - Opened by fenghuangzhige about 1 year ago
- 2 comments
Labels: stale
#16 - 请问基于中文语料库进行训练微调时有对原本的词表进行扩展吗
Issue -
State: closed - Opened by QJShan about 1 year ago
- 2 comments
Labels: stale
#15 - Can you first expose the training scripts, PT scripts and sft scripts, and synthesis scripts?
Issue -
State: closed - Opened by musellama about 1 year ago
- 3 comments
#14 - 请问预计什么时候开源代码和模型权重文件?
Issue -
State: closed - Opened by SunnyMarkLiu about 1 year ago
- 5 comments
#13 - about training efficient
Issue -
State: closed - Opened by 520jefferson about 1 year ago
- 2 comments
Labels: stale
#12 - Release v1.0: Chinese-LLaMA-2-7B, Chinese-Alpaca-2-7B
Pull Request -
State: closed - Opened by ymcui about 1 year ago
- 1 comment
#11 - 请教一下申请Lliama2模型下载Meta审批流程需要多久?
Issue -
State: closed - Opened by young-yang03 about 1 year ago
- 5 comments
#10 - 请教一下中文预训练数据组成?
Issue -
State: closed - Opened by brotherb about 1 year ago
- 2 comments
Labels: stale
#9 - 支援繁體中文嗎?
Issue -
State: closed - Opened by compustar about 1 year ago
- 5 comments
Labels: stale
#8 - 请问训练使用的显卡资源大概是多少
Issue -
State: closed - Opened by mafamily2496 about 1 year ago
- 2 comments
Labels: stale
#7 - 期待一下
Issue -
State: closed - Opened by RickyWang111 about 1 year ago
- 1 comment
Labels: stale
#6 - 期待作者早日放出模型,也欢迎使用试用我们自己尝试的模型
Issue -
State: closed - Opened by shiyemin about 1 year ago
- 1 comment