-
Notifications
You must be signed in to change notification settings - Fork 636
Insights: modelscope/ms-swift
Overview
Could not load contribution data
Please try again later
3 Releases published by 1 person
-
v3.3.0.post1
published
Apr 15, 2025 -
v3.3.1
published
Apr 26, 2025 -
v3.4.0
published
Apr 30, 2025
93 Pull requests merged by 11 people
-
[grpo] fix multi modal doc
#4124 merged
May 11, 2025 -
[grpo] support gen rm
#4151 merged
May 11, 2025 -
support internvl3 pretrain instruct
#4164 merged
May 11, 2025 -
[megatron]Support packing & CP
#4163 merged
May 11, 2025 -
Support ulysses streaming
#4160 merged
May 10, 2025 -
update readme
#4157 merged
May 9, 2025 -
Add more evaluation args
#4155 merged
May 9, 2025 -
Add sp script
#4154 merged
May 9, 2025 -
fix init parameters
#4148 merged
May 9, 2025 -
Fix bugs
#4150 merged
May 9, 2025 -
fix ulysses dpo
#4149 merged
May 9, 2025 -
Support init parameters
#4141 merged
May 9, 2025 -
Feature freezing/activating parameters via regex
#4143 merged
May 9, 2025 -
grpo code reward by judge0
#4140 merged
May 9, 2025 -
[megatron] support max_epochs
#4125 merged
May 9, 2025 -
[grpo] fix labels pop and peftmodel parameter check
#4136 merged
May 8, 2025 -
update qwen3 more models
#4123 merged
May 8, 2025 -
fix sequence_parallel
#4122 merged
May 7, 2025 -
fix omni aligner
#4117 merged
May 7, 2025 -
Fix ulysses eval
#4114 merged
May 7, 2025 -
fix packing
#4113 merged
May 7, 2025 -
fix enable_cache
#4109 merged
May 7, 2025 -
fix requirements
#4108 merged
May 7, 2025 -
[megatron] Update long text shell
#4106 merged
May 7, 2025 -
support max_epochs
#4102 merged
May 7, 2025 -
Update liger code
#4095 merged
May 6, 2025 -
fix enable_cache
#4091 merged
May 6, 2025 -
Support ulysses for llm/mllm,dpo/sft
#4085 merged
May 5, 2025 -
update docs
#4078 merged
May 4, 2025 -
feat: support megatron wandb
#4074 merged
May 4, 2025 -
feat: add run name support
#4072 merged
May 3, 2025 -
fix padding_side left
#4069 merged
May 3, 2025 -
support MiMo-7B
#4067 merged
May 2, 2025 -
fix packing eval streaming
#4066 merged
May 2, 2025 -
Support empty think loss scale
#4065 merged
May 2, 2025 -
support qwen3-moe awq
#4059 merged
May 1, 2025 -
Fix grpo eval when gas > 1
#4057 merged
May 1, 2025 -
fix rollout
#4055 merged
Apr 30, 2025 -
updates GRPOTrainer compatible with trl 0.17
#3969 merged
Apr 30, 2025 -
support Qwen2.5-Omni-3B
#4052 merged
Apr 30, 2025 -
update wechat
#4047 merged
Apr 30, 2025 -
Update readme & fix generate
#4041 merged
Apr 29, 2025 -
support qwen3_self_cognition
#4039 merged
Apr 29, 2025 -
fix grpo resume_from_checkpoint
#4035 merged
Apr 29, 2025 -
fix bugs
#4031 merged
Apr 28, 2025 -
Support Qwen3 series
#4029 merged
Apr 28, 2025 -
[Megatron] support MoE (Qwen2-Moe & Qwen3-MoE)
#4012 merged
Apr 28, 2025 -
fix truncation_strategy
#4025 merged
Apr 28, 2025 -
Fix gte training and compatible with ds3
#4022 merged
Apr 27, 2025 -
Fix merge sentence transformers
#4011 merged
Apr 27, 2025 -
[megatron] Support Qwen3
#3995 merged
Apr 26, 2025 -
Support vllm quantization
#4003 merged
Apr 26, 2025 -
🐛 fix: fix reward model train seq_cls
#3921 merged
Apr 26, 2025 -
fix seq_cls
#4002 merged
Apr 26, 2025 -
fix bugs
#4001 merged
Apr 26, 2025 -
fix get_toolcall & fix ci
#3999 merged
Apr 25, 2025 -
Fix web-ui
#3997 merged
Apr 25, 2025 -
Fix qwen2.5-omni use_audio_in_video
#3987 merged
Apr 24, 2025 -
Update unsloth compatibility
#3970 merged
Apr 24, 2025 -
fix parse tools
#3975 merged
Apr 24, 2025 -
Support hermes loss_scale
#3963 merged
Apr 23, 2025 -
fix bugs
#3962 merged
Apr 23, 2025 -
update docs
#3961 merged
Apr 23, 2025 -
Refactor Agent Template
#3918 merged
Apr 22, 2025 -
Decouple vLLM engine and GRPOTrainer.
#3911 merged
Apr 22, 2025 -
Support qwen3
#3945 merged
Apr 21, 2025 -
update qwen2_5_omni
#3908 merged
Apr 21, 2025 -
fix grpo doc
#3920 merged
Apr 17, 2025 -
revert swift_from_pretrained
#3914 merged
Apr 17, 2025 -
add rm center_rewards_coefficient argument
#3917 merged
Apr 17, 2025 -
Fix fp16 bf16
#3909 merged
Apr 17, 2025 -
Fix ui
#3903 merged
Apr 16, 2025 -
fix typealias to be compatible with Python 3.9
#3895 merged
Apr 16, 2025 -
fix bugs
#3893 merged
Apr 16, 2025 -
Fix glm4 z1
#3889 merged
Apr 15, 2025 -
Support kimi-vl
#3884 merged
Apr 15, 2025 -
refactor mm target_regex (compat peft/vllm)
#3879 merged
Apr 15, 2025 -
add paper link
#3886 merged
Apr 15, 2025 -
support glm4-z1
#3862 merged
Apr 15, 2025 -
fix grpo save checkpoint
#3865 merged
Apr 14, 2025 -
fix citest & minimax link
#3868 merged
Apr 14, 2025 -
Update swift docker
#3866 merged
Apr 14, 2025 -
support val_dataset_shuffle
#3860 merged
Apr 13, 2025 -
fix grpo completion length equal zero
#3857 merged
Apr 12, 2025 -
Update FAQ
#3841 merged
Apr 12, 2025 -
Fix multimodal target modules
#3858 merged
Apr 12, 2025 -
fix multimodal target_modules
#3856 merged
Apr 12, 2025 -
Fix internvl2.5/3 deepspeed packing
#3855 merged
Apr 12, 2025 -
support agent packing
#3853 merged
Apr 12, 2025 -
dapo-bug
#3846 merged
Apr 12, 2025 -
fix grpo filter overlong
#3844 merged
Apr 12, 2025 -
support internvl3
#3842 merged
Apr 12, 2025 -
Fix incorrect retry count check in LazyLLMDataset.__getitem__
#3845 merged
Apr 12, 2025
8 Pull requests opened by 5 people
-
Neptune completion logging
#3904 opened
Apr 16, 2025 -
feat: add LMDB support for multimodal resources
#3938 opened
Apr 19, 2025 -
fix enable_cache
#4075 opened
May 4, 2025 -
refactor grpo internal mode
#4097 opened
May 6, 2025 -
Refactor SP
#4121 opened
May 7, 2025 -
fix model_type mismatch
#4127 opened
May 8, 2025 -
support more vision dataset
#4132 opened
May 8, 2025 -
fix _tp_plan
#4167 opened
May 11, 2025
92 Issues closed by 44 people
-
Megatron SFT context_parallel_size>1时报cuda error
#4144 closed
May 11, 2025 -
pip install 'ms-swift[all]' -U的时候会进行很多个版本的下载
#4137 closed
May 9, 2025 -
Support for Qwen2-Audio and Qwen2.5-Omni
#4088 closed
May 8, 2025 -
qwen2.5-omni-7b merge-lora results differ
#3756 closed
May 8, 2025 -
raise IndexError(f"Index {index} out of range for dataset of size {size}.")
#4120 closed
May 8, 2025 -
Qwen2.5-7B-Base 超长文本训练部分step之后报错
#4105 closed
May 7, 2025 -
关于deepspeed多卡训练时.cache中出现和卡数成正比的数据拷贝，导致存储空间占用过大的问题
#3965 closed
May 6, 2025 -
Qwen3-8B-Base SFT 全参微调保存第一个模型后hang住
#4053 closed
May 6, 2025 -
Qwen3数据集设置不优雅
#4087 closed
May 6, 2025 -
Too many dataloader workers
#4061 closed
May 6, 2025 -
qwen3 seq_cls
#4073 closed
May 6, 2025 -
requirements中包的版本存在问题
#4080 closed
May 5, 2025 -
Support wandb logging in Swift Megatron SFT
#4071 closed
May 4, 2025 -
Add run_name argument support for wandb integration
#4046 closed
May 4, 2025 -
qwen2.5-vl推理时卡住
#3799 closed
May 3, 2025 -
是否计划支持XiaomiMiMo/MiMo-7B模型的微调？
#4064 closed
May 2, 2025 -
packing似乎和lazy_encode参数是冲突的？
#4054 closed
May 2, 2025 -
KTO使用自定义数据集报错
#4062 closed
May 2, 2025 -
grpo 训练卡住
#3887 closed
May 1, 2025 -
GRPO 课程学习
#3933 closed
May 1, 2025 -
grpo 设置vllm_server_host 报错
#3986 closed
May 1, 2025 -
3.3.1 版本 sft 微调internvl3 出现问题。
#4044 closed
Apr 30, 2025 -
3.4版本的sequence_parallel 被丢弃了吗？
#4043 closed
Apr 29, 2025 -
qwen2-vl 系列无法awq量化
#2649 closed
Apr 29, 2025 -
Lora 训练，merge 后，tokenizer.json 变大，推理生成乱码
#3502 closed
Apr 29, 2025 -
推理型大模型多轮对话SFT数据集构造问题请教
#3627 closed
Apr 29, 2025 -
能推理但是merge完部署后报错无法找到 adapter_config.json
#3643 closed
Apr 29, 2025 -
qwen2.5-7b-Instruct进行lora微调合并后推理报错
#3710 closed
Apr 29, 2025 -
swift训练时 Qwenvl2.0/2.5 是否采用了smart_resize
#3729 closed
Apr 29, 2025 -
Dataset preparation for Object Detection with Florence2
#1977 closed
Apr 29, 2025 -
npu 推理和部署怎么设置多卡
#2084 closed
Apr 29, 2025 -
qwen3 8B 全参数微调支持
#4037 closed
Apr 29, 2025 -
Resumming from checkpoint failed
#4032 closed
Apr 29, 2025 -
DeepSeek-R1-Distill-Qwen-1.5B这种模型该怎么准备SFT的数据？
#3996 closed
Apr 28, 2025 -
如何使用resume_from_checkpoint继续训练并增加数据集
#4000 closed
Apr 28, 2025 -
Possible bugs in MolmoE fintuning
#3998 closed
Apr 27, 2025 -
LoRA multi step training question
#4006 closed
Apr 27, 2025 -
set MAX_PIXELS=1003520 failed！
#4014 closed
Apr 27, 2025 -
Embedding模型训练到部署全流程
#4017 closed
Apr 27, 2025 -
Embedding模型训练到部署全流程
#4016 closed
Apr 27, 2025 -
regression训练的模型如何部署
#3786 closed
Apr 27, 2025 -
关于在GRPO训练配置reward model
#4010 closed
Apr 27, 2025 -
请教一个问题grpo 如何加入prm奖励，模型或者规则
#3952 closed
Apr 27, 2025 -
RuntimeError: CUDA error: CUDA-capable device(s) is/are busy or unavailable
#3960 closed
Apr 27, 2025 -
Inconsistency between `max_length` and `vllm_max_model_len` in official examples
#3956 closed
Apr 27, 2025 -
web ui模型导出，量化比特数需要增加一个None选项，导出模型必须量化不合理
#3994 closed
Apr 26, 2025 -
Pickling error: Mini-InternVL-Chat-2B-V1-5
#3303 closed
Apr 25, 2025 -
多标签分类训练数据问题请教
#3984 closed
Apr 25, 2025 -
grounding任务数据，一个<ref-object>同时出现在问题和回答中如何处理
#3979 closed
Apr 25, 2025 -
Too slow sft process
#3971 closed
Apr 24, 2025 -
自定义GRPO训练数据集加载失败
#3981 closed
Apr 24, 2025 -
推理结果不变，每次同样的图像，同样的问题，答案都一样。如何答案多样性？
#3978 closed
Apr 24, 2025 -
Is it possible to print inference samples during training via CLI?
#3968 closed
Apr 24, 2025 -
grounding 的损失计算
#3946 closed
Apr 23, 2025 -
Is there a specific method for training GRPO using Qwen2.5-VL-3B-Instruct with LoRA?
#3882 closed
Apr 23, 2025 -
grpo训练qwen2.5 7B 100steps后性能直线下降
#3875 closed
Apr 23, 2025 -
对InternVL3-8B进行微调时报错
#3959 closed
Apr 23, 2025 -
accuracy reward function in dapo
#3925 closed
Apr 23, 2025 -
grpo中的async模式是否能够支持tensor_parallel_size>1
#3712 closed
Apr 22, 2025 -
多机多卡情况下GRPO的异步训练问题
#3817 closed
Apr 22, 2025 -
请求添加sft 参数 fp16: false 透传到transformers 中的TrainingArguments，避免得到NAN
#3896 closed
Apr 22, 2025 -
InternVL3 lora 训练时解冻vit，freeze llm，训练新场景时，eval_acc 一直很低
#3890 closed
Apr 21, 2025 -
dpo全参训练错误，lora可正常训练
#3948 closed
Apr 21, 2025 -
Qwen 2.5 GPTQ int 4量化模型LORA微调报错
#3910 closed
Apr 21, 2025 -
qwn2.5-omni训练报错
#3905 closed
Apr 19, 2025 -
Confusion Regarding max_step
#3912 closed
Apr 18, 2025 -
奖励模型训练完模型结构不是打分模型
#3926 closed
Apr 18, 2025 -
使用qwen2.5-vl训练一个reward model，怎么在命令行中设置RewardConfig/RewardTrainer中需要的参数？
#3916 closed
Apr 17, 2025 -
qwen2.5vl 支持的 target module
#3900 closed
Apr 17, 2025 -
Target module not supported: 运行官方文档中Qwen2.5-VL Grounding推理脚本示例报错
#3913 closed
Apr 17, 2025 -
multi-node grpo training hangs
#3695 closed
Apr 17, 2025 -
Kimi-VL的模型支持
#3915 closed
Apr 17, 2025 -
输出logprobs会大幅度加大推理耗时
#3906 closed
Apr 17, 2025 -
AssertionError: quant_method: bnb, quantized model and does not support merge-lora.
#3859 closed
Apr 16, 2025 -
python3.9 is not compatible
#3894 closed
Apr 16, 2025 -
请问一下会支持kimi-vl的训练吗
#3869 closed
Apr 16, 2025 -
Qwen2.5-vl 怎么训练 DPO 模型？
#3861 closed
Apr 16, 2025 -
lora_llm_full_vit微调Qwen2.5VL时报错，前几天还是好的
#3891 closed
Apr 16, 2025 -
meet error when quantizing Qwen2.5vl-72B with multi-gpus
#3867 closed
Apr 16, 2025 -
DAPO train error
#3874 closed
Apr 14, 2025 -
InternVL3+DPO训练报错
#3870 closed
Apr 14, 2025 -
DS-Distill-Qwen7B GRPO:模型参数在多个地方共享?
#3873 closed
Apr 14, 2025 -
Qwen25VL 72B GRPO training (lora) would hang for no reason.
#3592 closed
Apr 14, 2025 -
NotImplementedError when merge lora for qwen2 vl 7B
#2344 closed
Apr 13, 2025 -
请问什么时候可以支持internvl3呢，我直接使用internvl模板会有问题吗
#3850 closed
Apr 13, 2025 -
windows平台lora训练qwen2vl
#3851 closed
Apr 12, 2025 -
确保结果的可复现性
#3767 closed
Apr 12, 2025 -
train的时候在val_dataset设置多个数据集问题
#3779 closed
Apr 12, 2025
151 Issues opened by 124 people
-
训练kimivl报错
#4166 opened
May 11, 2025 -
使用seq_cls功能loraa finetune qwenVL2.5inference时出现AttributeError: Identity has no attribute `weight
#4165 opened
May 11, 2025 -
grpo use speical token
#4162 opened
May 10, 2025 -
full sft设置了val_dataset后，在eval时报错
#4159 opened
May 10, 2025 -
Template _encode 函数内不能用model.cuda()
#4158 opened
May 9, 2025 -
swift sft 设置--streaming true时，会报 No such file or directory
#4156 opened
May 9, 2025 -
请问现在十分支持部署 基座qwen2.5-VL + 多个lora 这样的服务
#4153 opened
May 9, 2025 -
wandb，开了海外代理还一直报错（网络连接超时，network error (connectiontimeout)）
#4152 opened
May 9, 2025 -
使用megatron swift sft微调Qwen3-30B-A3B之后，checkpoint无法转回huggingface格式
#4147 opened
May 9, 2025 -
DPO训练效率很低
#4146 opened
May 9, 2025 -
Qwen2.5vl32B merge lora OOM问题
#4145 opened
May 9, 2025 -
dpo 是否支持packing
#4142 opened
May 9, 2025 -
可以实现不同数据使用不同的loss_scale吗
#4139 opened
May 8, 2025 -
自定义模型并注册,在数据map时卡住（版本3.3.1）
#4138 opened
May 8, 2025 -
Request Failed with 422 Error: Input Should Be a Valid String for Image Paths
#4135 opened
May 8, 2025 -
Some problems about loading Janus-Pro - traceback : Signal 11 (SIGSEGV) received by PID xxx
#4134 opened
May 8, 2025 -
swift megatron sys._base_executable problem
#4133 opened
May 8, 2025 -
在训练好的lora基础上用别的数据二次训练
#4131 opened
May 8, 2025 -
swift infer在tp=2的情况下，不支持deepseek-r1-distill-qwen系列和qwq32B模型的批推理
#4130 opened
May 8, 2025 -
swift infer的批处理非常好用，但能否支持近实时写入result_path，而不是最后写入
#4129 opened
May 8, 2025 -
Qwen2-audio-instruct用lora微调后inference，出现tensor维度不对应的问题
#4128 opened
May 8, 2025 -
支持Qwen3 MoE的Megatron LoRA训练
#4126 opened
May 8, 2025 -
raise IndexError(f"Index {index} out of range for dataset of size {size}.")
#4119 opened
May 7, 2025 -
GRPO下的多轮多模态对话数据集构建
#4118 opened
May 7, 2025 -
推理中出现从未遇见的bug
#4116 opened
May 7, 2025 -
有无懂哥说说internvl3_8B微调完后怎么做awq量化呀
#4115 opened
May 7, 2025 -
beta参数在GRPO中失效
#4112 opened
May 7, 2025 -
qwen omni注册的问题
#4110 opened
May 7, 2025 -
对于一个已经完成sft之后的任务，如果我想加入新的知识但不想掉点，我应该选择ms-swift实现的强化微调和GRPO哪个来完成呢？
#4107 opened
May 7, 2025 -
dpo模型RuntimeError: CUDA driver error: invalid argument，
#4104 opened
May 7, 2025 -
训练的时候总提示： RuntimeError: CUDA driver error: invalid argument
#4103 opened
May 7, 2025 -
LLama-omni进行audio微调索引报错
#4101 opened
May 7, 2025 -
ulysses raise NotImplementedError
#4100 opened
May 7, 2025 -
框架支持传rope theta的参数吗？
#4099 opened
May 6, 2025 -
序列分类模型在推理的时候会shuffle数据集
#4098 opened
May 6, 2025 -
internvl3_8B多模态模型的微调如何设置不同模块的冷冻与lora阶数呢？
#4096 opened
May 6, 2025 -
有什么参数可以调节dataset的sampling的比例
#4094 opened
May 6, 2025 -
sequence classification inference
#4093 opened
May 6, 2025 -
ModuleNotFoundError: No module named 'torch.distributed.device_mesh'
#4092 opened
May 6, 2025 -
可否在eval的过程中保存结果呢
#4090 opened
May 6, 2025 -
为啥现做RLHF 不支持sequence_parallel
#4089 opened
May 6, 2025 -
lora微调gte embedding, merge后推理结果跟微调的结果相差很大
#4084 opened
May 5, 2025 -
Streaming + Packing + resume_from_checkpoint时出现报错
#4083 opened
May 5, 2025 -
function call 微调报错 TypeError: string indices must be integers, not 'str'
#4082 opened
May 5, 2025 -
训练正常 eval时报assert error
#4081 opened
May 5, 2025 -
Pre-offline tokenize for ultra large multimodal datasets
#4079 opened
May 4, 2025 -
making llm_max_batch_size and mllm_max_batch_size configurable
#4077 opened
May 4, 2025 -
InternVL3-9B LoRA微调数据集预处理速度缓慢问题（大约7h）
#4076 opened
May 4, 2025 -
Fine-tuning Qwen2.5-Omni-7B with additional new layers on the audio tower
#4070 opened
May 3, 2025 -
how to run using vllm
#4068 opened
May 3, 2025 -
inference error with vllm 0.8.5
#4063 opened
May 1, 2025 -
用grpo训练qwen2.5-7b-instruct出现!!!!
#4060 opened
May 1, 2025 -
The expanded size of the tensor (8) must match the existing size (5) at non-singleton dimension 0.
#4056 opened
Apr 30, 2025 -
transformer_engine 安装失败
#4051 opened
Apr 30, 2025 -
dapo时在UserWarning: None of the inputs have requires_grad=True. Gradients will be None一直卡住，直至timeout
#4050 opened
Apr 30, 2025 -
dapo时在UserWarning: None of the inputs have requires_grad=True. Gradients will be None一直卡住，直至timeout
#4049 opened
Apr 30, 2025 -
DPO训练有支持长度惩罚的参数可选吗？
#4048 opened
Apr 30, 2025 -
[HELP]推理奖励模型报错，感谢大家，求教qwen基座rm后的模型如何vllm推理
#4045 opened
Apr 30, 2025 -
请求增加一个用于将通过ms-swift微调后的模型转为GGUF格式文件的Notebook文件
#4042 opened
Apr 29, 2025 -
如何将微调成功后的模型转为GGUF格式
#4040 opened
Apr 29, 2025 -
qwen3如何不使用思维链微调
#4038 opened
Apr 29, 2025 -
不支持bf16报错
#4036 opened
Apr 29, 2025 -
请求增加对Qwen3-8B的自我认知训练的NoteBook文件
#4034 opened
Apr 29, 2025 -
Resumming from checkpoint failed
#4033 opened
Apr 29, 2025 -
🚀 Best Practices for Training Qwen3/Qwen3-MoE
#4030 opened
Apr 28, 2025 -
qwen2_5VL missing apply_liger_kernel
#4028 opened
Apr 28, 2025 -
About the smart_resize of qwen2.5-vl
#4027 opened
Apr 28, 2025 -
3.4.0版本的swift会过滤数据集，是什么因素导致？
#4026 opened
Apr 28, 2025 -
请问有支持步骤奖励的rl方法么
#4024 opened
Apr 28, 2025 -
qwen2.5-vl-72b, vllm_server_host方式运行，CUDA out of memory
#4023 opened
Apr 28, 2025 -
qwen2.5-vl-72b多卡推理卡住
#4021 opened
Apr 27, 2025 -
关于系统提示词
#4020 opened
Apr 27, 2025 -
在新版本（3.4）中，如果nproc_per_node小于CUDA_VISIBLE_DEVICES的数量时无法运行，老版本（3.2）可以
#4019 opened
Apr 27, 2025 -
Embedding模型训练到部署全流程
#4018 opened
Apr 27, 2025 -
Inference speed between FP16 and F16
#4015 opened
Apr 27, 2025 -
尝试使用CPU训练时，无法将任务分布到多CPU上
#4013 opened
Apr 27, 2025 -
swift微调Qwen2.5-VL，合并模型后，分别使用swift和transform推理结果不一致
#4009 opened
Apr 27, 2025 -
About the support for Kimi-Audio
#4008 opened
Apr 27, 2025 -
关于qLoRA训练
#4007 opened
Apr 27, 2025 -
原始gte 7B 模型大小大概29G， 使用github，训练脚本使用example中对应的训练参数，改为全参训练，参数变成 14G。GTE模型全参训练完加载报错
#4005 opened
Apr 27, 2025 -
model difference
#4004 opened
Apr 26, 2025 -
Bug! After resuming training, the info line doesn't show details anymore ...
#3993 opened
Apr 25, 2025 -
infer速度远远小于train期间infer val数据集的速度
#3992 opened
Apr 25, 2025 -
deepspeed报错
#3991 opened
Apr 25, 2025 -
端口监听错误
#3988 opened
Apr 24, 2025 -
使用Qwen2.5-VL-72B推理时shutdown
#3985 opened
Apr 24, 2025 -
Megatron 近期是否有计划支持 vlm
#3982 opened
Apr 24, 2025 -
Training data in a sequential order, without shuffle
#3977 opened
Apr 24, 2025 -
ms-swift-3.4.0.dev0 推理seq_cls(文本序列分类)未能输出logprobs
#3976 opened
Apr 24, 2025 -
微调DS_32B后merge_lora，将合并后的模型推理不生效
#3974 opened
Apr 24, 2025 -
swift web-ui 显卡编号映射异常
#3973 opened
Apr 24, 2025 -
Gemma3 Pan and Scan breaks tokenization
#3972 opened
Apr 24, 2025 -
在inference的时候指定--max_length 4096但是似乎没有起到任何作用
#3967 opened
Apr 23, 2025 -
gte类的模型 inf-v1-1.5b embedding模型训练问题
#3966 opened
Apr 23, 2025 -
奇怪的out of memory报错
#3964 opened
Apr 23, 2025 -
swift app
#3958 opened
Apr 22, 2025 -
Evaluation result of MMMU_DEV_VAL cannot be reproduced on Swift
#3957 opened
Apr 22, 2025 -
KTO训练，1张卡没问题，8张卡出现问题：Tensor Tensor dtypes: BFloat16vs Float
#3955 opened
Apr 22, 2025 -
steps如何计算的
#3954 opened
Apr 22, 2025 -
关于 DeepSpeed Config 的问题
#3953 opened
Apr 22, 2025 -
关于断点续训的问题resume_from_checkpoint
#3951 opened
Apr 21, 2025 -
怎么在infer的时候传system prompt？
#3950 opened
Apr 21, 2025 -
GPTQ量化模型GRPO强化微调报错:AttributeError: 'GPTQLoraLinear' object has no attribute 'get_delta_weight'
#3949 opened
Apr 21, 2025 -
使用ToolBench数据集出错
#3947 opened
Apr 21, 2025 -
微调Intervl2-8B后，使用webui推理一直无输出，且没报错
#3944 opened
Apr 21, 2025 -
if sleep_level > 0, gradient_accumulation_steps will be forced to 1
#3943 opened
Apr 21, 2025 -
Qwen2.5 VL dataset sampler based on modality length
#3942 opened
Apr 20, 2025 -
Tensorboard to `--bind_all`
#3941 opened
Apr 20, 2025 -
自定义数据集中，有的样本有图片，有的样本没有图片，该如何处理呢？
#3940 opened
Apr 20, 2025 -
dataset中指定多个数据集，图文混合训练的问题
#3939 opened
Apr 19, 2025 -
请教怎么使用swift infer
#3936 opened
Apr 18, 2025 -
UserWarning: None of the inputs have requires_grad=True
#3935 opened
Apr 18, 2025 -
The GRPO training process hangs for multi-node training.
#3934 opened
Apr 18, 2025 -
QWQ：GRPO训练无法跑通，报错”RuntimeError: ACL stream synchronize failed, error code:107020“
#3932 opened
Apr 18, 2025 -
在GRPO训练中Weight_decay似乎没奏效？
#3931 opened
Apr 18, 2025 -
2张3090训练70B模型的脚本为啥是7B的模型
#3929 opened
Apr 18, 2025 -
LLava 跑GRPO 无法跑通
#3928 opened
Apr 18, 2025 -
使用 AutoModelForSequenceClassification 训练 seq_cls 任务时出错
#3927 opened
Apr 18, 2025 -
Qwen2VL微调到某个step就报错
#3924 opened
Apr 18, 2025 -
swift infer 单机多卡显存分配不均
#3923 opened
Apr 17, 2025 -
support Qwen 3 and Qwen 3 MoE
#3922 opened
Apr 17, 2025 -
VLLM 运行GLM-4-32B-0414的推理需要的配置？
#3919 opened
Apr 17, 2025 -
[Bug]: Wrong context length for Qwen 2.5 7B-Instruct?
#3907 opened
Apr 17, 2025 -
ms-swfit 微调qewn2.5-vl 报错 IndexError: index 4247 is out of bounds for dimension 0 with size 4247
#3902 opened
Apr 16, 2025 -
ms-swift vs r1v in grpo of Qwen-2.5vl
#3901 opened
Apr 16, 2025 -
对于自定义的数据集，如何计算其token数量？
#3899 opened
Apr 16, 2025 -
希望增加qwen2.5-omni推理保存音频文件
#3898 opened
Apr 16, 2025 -
ImportError: cannot import name 'Qwen2_5OmniModel from 'transformers'
#3897 opened
Apr 16, 2025 -
Do swift support these type of data for training multimodal reward model（having value head）
#3888 opened
Apr 15, 2025 -
AttributeError: 'NoneType' object has no attribute 'shape'
#3885 opened
Apr 15, 2025 -
微调qwen2.5-vl做点检测的grounding，数据集应该是什么形式
#3883 opened
Apr 15, 2025 -
merge_lora.sh出错
#3881 opened
Apr 15, 2025 -
DPO训练log打印日志:logits/chosen和logits/rejected完全一样
#3880 opened
Apr 15, 2025 -
There seems not to be a single sample in your epoch_iterator
#3878 opened
Apr 15, 2025 -
grpo log bug
#3877 opened
Apr 15, 2025 -
GRPO 训练100 steps后性能骤降，请问是什么原因
#3876 opened
Apr 15, 2025 -
VAPO支持计划
#3872 opened
Apr 14, 2025 -
grpo训练32b模型OOM
#3871 opened
Apr 14, 2025 -
GRPO Example script results
#3852 opened
Apr 12, 2025 -
单张4090对minicpmV2.6进行视频问答微调总是中途OOM
#3849 opened
Apr 12, 2025 -
Meet GPU OutOfMemory in GRPO training
#3848 opened
Apr 12, 2025
42 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.
-
grpo TypeError: CosineReward.__call__() missing 1 required positional argument: 'solution'
#3840 commented on
Apr 12, 2025 • 0 new comments -
LISA训练要么OOM，要么使用deepseed就报错
#2035 commented on
Apr 12, 2025 • 0 new comments -
请问目前使用deepspeed进行 sft 支持 pipe parallel 和 tensor parallel 配置么，是否考虑支持一下？
#2537 commented on
Apr 12, 2025 • 0 new comments -
Qwen2-VL-7B-Instruct微调爆显存
#3405 commented on
Apr 13, 2025 • 0 new comments -
qwen2.5 3b sft 多机训练 运行到固定的某个step会有一个节点显存 OOM
#3327 commented on
Apr 14, 2025 • 0 new comments -
多GPU时，一开use_liger就报错，
#2261 commented on
Apr 16, 2025 • 0 new comments -
swift3.1.0.dev0在推理Florence2时错误
#3133 commented on
Apr 17, 2025 • 0 new comments -
Qwen2.5-VL训练train memory逐步升高
#3741 commented on
Apr 18, 2025 • 0 new comments -
导出模型后，再导入ollama，一直重复回答不停止
#3167 commented on
Apr 18, 2025 • 0 new comments -
Loss goes to 0, Gibberish Outputs
#3582 commented on
Apr 20, 2025 • 0 new comments -
qwen 2.5 vl 72B A100单机rm训练爆内存，不是显存，是不是哪里内存没回收？
#3559 commented on
Apr 21, 2025 • 0 new comments -
求一个能8卡A100使用GRPO跑通Qwen2.5 72B模型的脚本
#3416 commented on
Apr 21, 2025 • 0 new comments -
如何微调自有的COT数据集
#3535 commented on
Apr 22, 2025 • 0 new comments -
评测参数bug
#3770 commented on
Apr 22, 2025 • 0 new comments -
训练一半显存爆了
#3805 commented on
Apr 22, 2025 • 0 new comments -
qwen2.5 omni TypeError: 'NoneType' object is not iterable
#3715 commented on
Apr 23, 2025 • 0 new comments -
Any plans to support megatron for GRPO training?
#3760 commented on
Apr 26, 2025 • 0 new comments -
unsloth error when sft qwen2.5-vl-7b-instruct
#3682 commented on
Apr 26, 2025 • 0 new comments -
grpo训练卡住，一直显示一下问题。
#3794 commented on
Apr 27, 2025 • 0 new comments -
序列并行批次维度不匹配
#3014 commented on
Apr 28, 2025 • 0 new comments -
llama-3.2-3b instruct doesn't stop writing
#2184 commented on
Apr 28, 2025 • 0 new comments -
qwen2-vl 的 pretrain 是否支持
#2222 commented on
Apr 29, 2025 • 0 new comments -
强化学习训练MLLM
#2212 commented on
Apr 29, 2025 • 0 new comments -
ms-swift3 Suggestion Box
#2217 commented on
Apr 29, 2025 • 0 new comments -
Megatron-SWIFT训练导出32B模型显存报错
#3768 commented on
Apr 30, 2025 • 0 new comments -
请问如何在grpo中配置自定义的数据集路径，并进行数据格式转换？
#3525 commented on
May 1, 2025 • 0 new comments -
NPU训练qwen2.5-vl报错
#3408 commented on
May 2, 2025 • 0 new comments -
GRPO Training Speed Testing
#3302 commented on
May 2, 2025 • 0 new comments -
微调了qwen2-audio-7b-instruct
#2637 commented on
May 5, 2025 • 0 new comments -
QwenVL2 72B 序列并行报错维度不匹配
#2972 commented on
May 5, 2025 • 0 new comments -
训练中途突然报错 NCCL watchdog thread terminated with exception
#1817 commented on
May 6, 2025 • 0 new comments -
Customized Image Data Augmentation
#2345 commented on
May 7, 2025 • 0 new comments -
cannot import name 'LoRA' from 'swift'
#3665 commented on
May 7, 2025 • 0 new comments -
lora微调后再awq量化，报错， 详细如下：
#2318 commented on
May 7, 2025 • 0 new comments -
Qwen2.5-vl 微调grounding任务，怎么使用自己本地数据集训练
#3204 commented on
May 8, 2025 • 0 new comments -
请求支持健康检查
#3474 commented on
May 8, 2025 • 0 new comments -
SimPO and ORPO support for VLM (Qwen2.5VL)
#3718 commented on
May 8, 2025 • 0 new comments -
多卡多进程使用orpo卡死，触发watchdog caught collective operation timeout.
#3564 commented on
May 8, 2025 • 0 new comments -
支持GME微调么
#3019 commented on
May 10, 2025 • 0 new comments -
grpo liger loss
#3781 commented on
May 1, 2025 • 0 new comments -
[WIP] support sglang engine
#3810 commented on
Apr 26, 2025 • 0 new comments