AIDC-AI/Ovis

Issues

• Grounding Ability

  #43 opened 14 days ago by zizaisuiyuan
  0

• How to correctly input image-free prompt?

  #37 opened 16 days ago by HaiFangfang
  1

• Visual Tokenizer: IMAGE_ATOM_ID

  #42 opened 17 days ago by KyujinHan
  0

• Traning script and data_info_v1.6 will be open source?

  #23 opened 3 months ago by liuheng0111
  3

• A few questions about `train.py`

  #41 opened 22 days ago by KyujinHan
  0

• how to create a custom dataset ?

  #40 opened 25 days ago by himasai9712
  0

• Why use different dataset for Training Ovis1.5-Gemma2-9B-S3 and Ovis1.5-Llama3-8B-S3

  #39 opened a month ago by LIRENDA621
  0

• Are you planning to release a small version for edge devices like Jetson AGX Orin?

  #38 opened a month ago by ALFONSOBUGRA
  0

• When will the 27B Gemma version model be released?

  #34 opened a month ago by ALFONSOBUGRA
  2

• Any plan about support video and audio?

  #35 opened 2 months ago by thesby
  1

• Ovis1.6-Llama3.2-3B-GPTQ-Int4. How can it be inferred using the CPU?

  #36 opened 2 months ago by peterlong2003
  1

• Finetune Ovis 1.6 with LoRA

  #27 opened 3 months ago by anhquannguyen21
  2

• running on 4bit model

  #29 opened 3 months ago by haiderasad
  2

• Can I use this model with the CPU？

  #32 opened 2 months ago by peterlong2003
  0

• How to use multiple GPUs for inference?

  #12 opened 4 months ago by waltonfuture
  8

• ValueError: Model architectures ['Ovis'] are not supported for now.

  #33 opened 2 months ago by yangxin60-tal
  0

• If it is possible to run inference with OVIS 1.6 on a single 4090 GPU?

  #22 opened 3 months ago by Raven625
  6

• Run into 'HybridCache' object has no attribute 'max_batch_size' error when doing inference

  #31 opened 3 months ago by dustinjoe
  3

• 内存不断增加, 内存泄漏

  #30 opened 3 months ago by JianbangZ
  1

• Missing datasets

  #28 opened 3 months ago by Hutaf
  0

• Can I identify and analyze videos? How to input video? Do you have any examples

  #25 opened 3 months ago by libai-lab
  1

• About the equivalence and a slightly more complex MLP connection

  #5 opened 5 months ago by lucasjinreal
  12

• The size of tensor a (3084) must match the size of tensor b (3085) at non-singleton dimension 3

  #18 opened 3 months ago by longkeyy
  1

• training script

  #20 opened 3 months ago by ecnuycxie
  2

• Failed to process batch: Currently, only support `batch_size=1`

  #19 opened 3 months ago by Phoenix724
  5

• 支持输入多张image吗，支持的话能给个示例吗

  #24 opened 3 months ago by yangxin60-tal
  1

• What does IMAGE_INDICATOR_IDS mean?

  #21 opened 3 months ago by SamaelChen
  1

• about license

  #17 opened 3 months ago by ecnuycxie
  2

• The training effect is poor using official data and code

  #15 opened 3 months ago by liuheng0111
  1

• How to finetune the model？

  #16 opened 3 months ago by cyj95
  1

• > We have uploaded the images for `ai2d-mc-15k` to [Huggingface](https://huggingface.co/datasets/AIDC-AI/Ovis-dataset/blob/main/images/ai2d-mc-15k.zip).

  #11 opened 3 months ago by liuheng0111
  1

• can not find ocr-469k images

  #13 opened 4 months ago by luyao-cv
  1

• hwl-eng-10k 的jsonl文件是哪个呀？

  #14 opened 4 months ago by luyao-cv
  1

• About the dataset

  #10 opened 4 months ago by lucasjinreal
  1

• Some dataset can not download

  #9 opened 4 months ago by liuheng0111
  2

• Is SoftMax Computation-Intensive?

  #8 opened 4 months ago by wusize
  1

• 有没有完整的数据集

  #7 opened 5 months ago by liuheng0111
  2

• The training strategy in the Ablation Study

  #6 opened 5 months ago by ruxin123
  1

• 数据集访问不到

  #1 opened 5 months ago by liuheng0111
  5

• Missing *.parquet files in https://huggingface.co/datasets/AIDC-AI/Ovis-dataset/tree/main/meta_files/v1_5

  #3 opened 5 months ago by SeuTao
  1

• Prompts used to build In-house data

  #2 opened 5 months ago by zjy-ucas
  2

• Missing metadata files in v1.5

  #4 opened 5 months ago by kakao-logan-c
  2