Pinned issues
Issues
- 4
运行llava-llama-3-8b-v1_1报错:FileNotFoundError: can't find *_optim_states.pt files in directory
#674 opened by Egber1t - 2
- 3
past_length is None
#722 opened by vincent507cpu - 2
精调停止在这里,什么问题?
#718 opened by chalesguo - 2
- 0
- 0
question on training on multi gpus
#725 opened by ztfmars - 2
- 0
error while finetune llava_llama3_8b_instruct_qlora_clip_vit_large_p14_336_e1_gpu1_finetune_copy.py on my own data
#723 opened by J0eky - 0
指定模型本地路径后仍然需要下载模型
#721 opened by vincent507cpu - 1
Execution exits unexpectedly: llava_llama3_8b_instruct_qlora_clip_vit_large_p14_336_e1_gpu1_finetune
#717 opened by chalesguo - 1
执行 NPROC_PER_NODE=2 xtuner train /root/StableDiffusionGPT/config/internlm2_1_8b_qlora_alpaca_e3_copy.py --work-dir /root/test/ft/train --deepspeed deepspeed_zero2 指令运行报错
#719 opened by LTtt456c - 1
what's purpose of dispatch
#716 opened by shockjiang - 1
目前支持多模态LLM的微调吗,比如qwen-vl、internLM-VL等
#720 opened by ybshaw - 2
- 2
introduce cogvlm2
#714 opened by Jayantverma2 - 8
xtuner转化模型权重文件是否支持多卡
#715 opened by AlittlePIE - 1
支持imagebind多模态大模型微调或者训练吗?
#711 opened by lewis-ing - 1
could you provide a internlm2 1.8B llava model
#712 opened by shockjiang - 16
- 1
华为Ascend NPU卡训练的时候,需要设置哪些参数?
#709 opened by apachemycat - 1
- 2
ERROR: Could not find a version that satisfies the requirement mpi4py-mpich (from xtuner[all]) (from versions: none)
#704 opened by eugine5 - 9
- 0
请问在32k的长度下微调7B模型,大概需要多少显存呢?
#707 opened by hxujal - 10
支持Yi-1.5系列Chat模型
#698 opened by thomas-yanxin - 1
llava-llama-3-8b-v1_1-hf How to quantify awq?
#699 opened by goodnight654 - 5
合并llama3时出现如下报错,这个问题再使用zero3时也出现了
#686 opened by 1518630367 - 0
Plans to support InternLM-XComposer
#705 opened by babla9 - 0
使用xtuner进行LLaVA训练如何混合纯文本数据
#701 opened by thomas-yanxin - 1
关于多轮对话loss mask计算的疑问
#700 opened by RyanOvO - 2
数据在入过程中样本量减少
#682 opened by Jason8Kang - 3
The sequence parallel is open when I don't use it.
#669 opened by amulil - 3
请问chatglm3的lora微调需要多大显存呀
#694 opened by Franklin-L - 2
求助max_epochs 的问题
#677 opened by Franklin-L - 3
log输出中的time的意思
#687 opened by shockjiang - 2
支持自定义视觉编码器么(llava-llama3)?
#668 opened by Yanllan - 2
any support plan for llava-llama3-70b/12b? any guide for Module optimization ?
#689 opened by ztfmars - 1
无法启动训练,似乎是mmengine有问题
#691 opened by Dominic23331 - 1
warning on CUTLASS&sparse_attn&triton
#676 opened by shockjiang - 1
- 2
使用官方脚本对应数据集提示列名不匹配,使用同格式自定义数据集报错
#692 opened by LumenScope - 2
shape mismatch when loading llava-phi path
#681 opened by shockjiang - 1
报错FileNotFoundError: [Errno 2] No such file or directory: '/app/work_dirs/chatglm2_6b_qlora_lawyer_e3_copy/20240514_035914/vis_data/eval_outputs_iter_499.txt'
#685 opened by rcejzibjks38 - 2
如何再8*A100上预训练128k长度的llama3?
#683 opened by 1518630367 - 0
safetensors文件权重
#680 opened by WEXIJUE - 0
support flash-attn in Phi3
#679 opened by shockjiang - 5
LLaVAModel finetuning with llava-phi-3-mini-xtuner
#671 opened by shockjiang - 0
- 1
xtuner mmbench on xtuner/ lava-Llama 3-8b-v1_1-hf
#673 opened by Yanllan