-
Notifications
You must be signed in to change notification settings - Fork 418
Insights: modelscope/ms-swift
Overview
Could not load contribution data
Please try again later
1 Release published by 1 person
-
v3.0.0
published
Dec 23, 2024
31 Pull requests merged by 3 people
-
fix alpaca
#2771 merged
Dec 26, 2024 -
support modern_bert & support bert deploy
#2767 merged
Dec 26, 2024 -
fix app-ui
#2765 merged
Dec 25, 2024 -
fix shell
#2764 merged
Dec 25, 2024 -
fix bugs
#2761 merged
Dec 25, 2024 -
fix web-ui
#2758 merged
Dec 25, 2024 -
support SequenceClassification & update QVQ-72B-Preview
#2747 merged
Dec 24, 2024 -
fix docs multimodal
#2742 merged
Dec 24, 2024 -
Fix windows encoding gbk
#2741 merged
Dec 24, 2024 -
support AI-ModelScope/Skywork-o1-Open-Llama-3.1-8B
#2739 merged
Dec 23, 2024 -
support multi-modal llamapro
#2738 merged
Dec 23, 2024 -
fix windows
#2733 merged
Dec 23, 2024 -
support paligemma2
#2735 merged
Dec 23, 2024 -
remove files
#2732 merged
Dec 23, 2024 -
update examples
#2730 merged
Dec 23, 2024 -
support iic/DocOwl2
#2728 merged
Dec 23, 2024 -
Support more internvl2.5 awq/mpo & internvl2 pretrain model
#2726 merged
Dec 22, 2024 -
Support qwen agent format
#2722 merged
Dec 21, 2024 -
fix batch_infer pad_token & florence
#2725 merged
Dec 21, 2024 -
Fix mplug owl2, molmo
#2724 merged
Dec 21, 2024 -
Better error messages
#2721 merged
Dec 20, 2024 -
fix gptq group_size
#2720 merged
Dec 20, 2024 -
fix examples
#2719 merged
Dec 20, 2024 -
fix deploy request_config
#2718 merged
Dec 20, 2024 -
update examples
#2714 merged
Dec 20, 2024 -
support Qwen/QVQ-72B-Preview
#2712 merged
Dec 19, 2024 -
Fix multi lora
#2711 merged
Dec 19, 2024 -
fix timeout & web-ui
#2709 merged
Dec 19, 2024 -
qwen to Qwen
#2708 merged
Dec 19, 2024 -
Update FAQ
#2706 merged
Dec 19, 2024 -
fix eval strategy
#2707 merged
Dec 19, 2024
1 Pull request opened by 1 person
-
add 'right' option for 'truncation_strategy'
#2754 opened
Dec 24, 2024
24 Issues closed by 8 people
-
关于Alpaca型数据处理方式的疑问
#2766 closed
Dec 26, 2024 -
ValueError: Please set --model <model_id_or_path>`, model: None
#2770 closed
Dec 26, 2024 -
swift 3.0 web-ui deployment error
#2755 closed
Dec 25, 2024 -
pretrain的数据结构问题
#2750 closed
Dec 24, 2024 -
ms-swift 3 or 3.1 版本训练qwen2-vl-2b-instruct时，不能同时启用deepspeed和flash-attn
#2751 closed
Dec 24, 2024 -
minicpmv-2.6微调CUDA out of memory.
#2731 closed
Dec 24, 2024 -
[WARNING:swift] Current length of row(2210) is larger than the max_length(2048), deleted.
#2501 closed
Dec 23, 2024 -
SFT之后的模型再进行fine tune？
#2521 closed
Dec 23, 2024 -
指定cuda device无效
#2562 closed
Dec 23, 2024 -
InternVL推理 KeyError: 'Qwen2ForCausalLM'
#2566 closed
Dec 23, 2024 -
3.0版本的说明文档可读性很差
#2567 closed
Dec 23, 2024 -
What is the difference between the prompt template of Qwen2-VL and Qwen2-VL-Instruct?
#2578 closed
Dec 23, 2024 -
请问有internvl2.5单视频推理脚本吗？
#2620 closed
Dec 23, 2024 -
请问v3.0怎么添加自己的训练数据，dataset_info.json和2.x版本变化挺大的
#2642 closed
Dec 23, 2024 -
多机多卡场景下resume_from_checkpoint，报错assert len(self.ckpt_list) > 0
#2644 closed
Dec 23, 2024 -
ValueError: model_type: 'internvl2_5-38b' is not registered.
#2676 closed
Dec 23, 2024 -
cannot import name 'run_deploy' from 'swift.llm'
#2653 closed
Dec 23, 2024 -
当--predict_with_generate True进行SFT时，出现下面的错误，swift版本：2.6.1
#2672 closed
Dec 23, 2024 -
PaliGemma 2视觉语言模型支持
#2695 closed
Dec 23, 2024 -
qwen2.5-72b微调后如何部署到Xinference
#2734 closed
Dec 23, 2024 -
autoawq failed: unexpected keyword argument 'use_cache'
#2729 closed
Dec 23, 2024 -
group_size is not passed to quantizer while export model with gptq quant
#2710 closed
Dec 20, 2024 -
推理部署服务的api文档
#2684 closed
Dec 19, 2024 -
evaluation_strategy问题
#2704 closed
Dec 19, 2024
25 Issues opened by 22 people
-
请问下断点重训具体命令是什么啊
#2768 opened
Dec 25, 2024 -
DeepSeek-VL2推理报错
#2763 opened
Dec 25, 2024 -
Any plan for the support of MPO training
#2762 opened
Dec 25, 2024 -
swift3 internvl2_5 双机16卡，lora 单卡OOM
#2760 opened
Dec 25, 2024 -
internvl 2.5 gptq 量化失败
#2759 opened
Dec 25, 2024 -
qwen2-vl-7b爆内存，注意不是显存，是爆内存！内存没回收= =
#2757 opened
Dec 24, 2024 -
使用swift lmdeploy 进行模型推理时，随着推理次数增加，物理内存会撑爆
#2756 opened
Dec 24, 2024 -
qwen2-vl-7b-instruct out of memory
#2753 opened
Dec 24, 2024 -
swift2.+版本vllm推理qwen2.5 模型gptq-int4版本报错
#2752 opened
Dec 24, 2024 -
swift训练reward model报错
#2749 opened
Dec 24, 2024 -
qwen2 vl 用 swift deploy 做部署如何进行视频问答？
#2748 opened
Dec 24, 2024 -
MaxLengthError for pre-training
#2745 opened
Dec 24, 2024 -
Qwen2-Audio SFT 报错
#2744 opened
Dec 24, 2024 -
qwen2-vl-7b video full sft OOM
#2743 opened
Dec 24, 2024 -
是否有GPTQModel支持的计划
#2740 opened
Dec 23, 2024 -
fsdp + qlora 单机多卡运行失败
#2737 opened
Dec 23, 2024 -
swift eval中评测微调lora后的模型，openai._base_client显示retrying request
#2736 opened
Dec 23, 2024 -
KTO训练，evaluation阶段报错
#2727 opened
Dec 22, 2024 -
自定义插件优化器对分组参数设置不同的学习率，在使用deepspeed后无效
#2723 opened
Dec 20, 2024 -
KTO训练，使用modelscope预料，训练loss没有任何变化
#2717 opened
Dec 20, 2024 -
微调glmglm-4-9b-chat-hf 时报 “KeyError: 'auto_map'”错误
#2716 opened
Dec 20, 2024 -
单服务器双4090微调qwen2.5-14B-Instruct成功，多机多卡2台服务器4张4090微调启动失败，CUDA out of memory
#2715 opened
Dec 20, 2024 -
kto训练：ValueError: remaining_argv: ['--train_type', 'sft']
#2713 opened
Dec 19, 2024
17 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.
-
能否集成一些自动化调参工具，并且提供一下3.0的pip包，现在download还是2.6.1版本的包。
#2705 commented on
Dec 19, 2024 • 0 new comments -
merge multi loras
#2702 commented on
Dec 19, 2024 • 0 new comments -
多模态模型多轮对话训练数据格式
#2673 commented on
Dec 20, 2024 • 0 new comments -
BUG: 训练集中包含特殊字符<unk>时有报错
#2688 commented on
Dec 23, 2024 • 0 new comments -
3.0 --device_max_memory
#2655 commented on
Dec 23, 2024 • 0 new comments -
Full SFT 的EMA支持
#2685 commented on
Dec 23, 2024 • 0 new comments -
v2.3.2以后InternVL2-26B训练比之前慢15%左右
#2502 commented on
Dec 23, 2024 • 0 new comments -
GOT-OCR 2.0 训练新的prompt数据后发现推理没有任何效果
#2259 commented on
Dec 24, 2024 • 0 new comments -
qwen2-vl 系列无法awq量化
#2649 commented on
Dec 24, 2024 • 0 new comments -
Best Practices for Inference and Fine-Tuning with MiniCPM-V 2.6
#1613 commented on
Dec 24, 2024 • 0 new comments -
pretrain报错进度异常问题
#2692 commented on
Dec 24, 2024 • 0 new comments -
Best practice for Qwen2-Audio
#1653 commented on
Dec 24, 2024 • 0 new comments -
lora 微调的模型使用--resume_from_checkpoint参数，继续训练报显存不足；不使用--resume_from_checkpoint参数可以正常训练
#2505 commented on
Dec 25, 2024 • 0 new comments -
Visualization of Grounding Tasks
#2635 commented on
Dec 25, 2024 • 0 new comments -
mplug-owl3-7b-chat fine-tuning document
#1969 commented on
Dec 25, 2024 • 0 new comments -
qwen2-vl 的 pretrain 是否支持
#2222 commented on
Dec 26, 2024 • 0 new comments -
ms-swift3 Suggestion Box
#2217 commented on
Dec 26, 2024 • 0 new comments