Ecosyste.ms: Timeline
Browse the timeline of events for every public repo on GitHub. Data updated hourly from GH Archive.
zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
通过“I2V”模型进行图片到视频推理出现大量噪声
### System Info / 系統信息 使用cli_demo.py文件 通过“I2V”模型进行图片到视频推理出现大量噪声 ### Information / 问题信息 - [X] The official example scripts / 官方的示例脚本 - [ ] My own modified scripts / 我自己修改的脚本和任务 ### Reproduction...
zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
Using gradio_web_demo.py script to inference t2v model is very slowly
### System Info / 系統信息 ubuntu 22.0 ### Information / 问题信息 - [X] The official example scripts / 官方的示例脚本 - [ ] My own modified scripts / 我自己修改的脚本和任务 ### Reproduction / 复现过程 python gradio_web_dem...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
The translation of the given text is: "We can refer to the community's approach to implement V2V, and the T2V model can be modified. But we haven't tried I2V."
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
emm, num_frames needs to be changed to 161, export_to_video(video, output_path, fps=16). Also, have you tried whether five seconds is normal.
zRzRzRzRzRzRzR created a comment on an issue on THUDM/GLM-Edge
结构上有一定的区别，但是差距不大，主要目的是在模型变小后还能保持一定的性能考虑
zRzRzRzRzRzRzR closed an issue on THUDM/GLM-Edge
(Suggestion) Add EOS after each message turn in your future models
### Feature request / 功能建议 Hello, thanks for releasing this model. I'm Son from Hugging Face and I'm one of the contributor of llama.cpp I would like to contact you regarding the subject ...
zRzRzRzRzRzRzR pushed 2 commits to main THUDM/GLM-Edge
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
By adding noise to images during the training process, the model is able to learn how to handle noisy data. This data augmentation strategy can enhance the model's adaptability to different input d...
zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
1.5-I2V训练首帧复制问题
非常感谢您的工作！ 我在1.5I2V版本的推理代码中发现，noise latents是22帧，然后image latents在第0帧处与noise在特征维度拼接。 想确认一下，在训练过程中的操作，是否是将81帧的视频经过vae编码为21帧，复制首帧放在最前面得到22帧（即第0和第1帧latents是相同的？），然后image latents拼接在0帧处？计算loss的时候是使用22帧一起...
zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
The conditional image does not seem to be effective in your image-to-video fine-tuning
Thank you for your excellent work! I have a question about your image-to-video finetuning. I think you adopt method [encode_video](https://github.com/THUDM/CogVideo/blob/main/finetune/train_cogv...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
The official diffusers fine-tuning has not implemented the t patch, as you have observed. This is because we are currently using fine-tuning in cogvieox-factory, so the script here is for cogvideox...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
Change the frame rate from fps=8 to fps=16 in the export_to_video function Nevertheless, in this scenario and with this prompt word, the effect is indeed not arbitrary, and this issue is recorde...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
There are several questions 1. Check if your model is freshly cloned from Hugging Face, and whether the sample height and width are 300. 2. Whether to install diffusers from source code 3. Onl...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/GLM-Edge
Our model group colleagues said they would consider this plan for the next generation model!
zRzRzRzRzRzRzR closed an issue on THUDM/GLM-Edge
Reuqest: QNN Qualcomm inference example
### Feature request / 功能建议 Hi all, the README mentions testing on the Qualcomm 8 Elite (Gen4) platform with all models running on the NPU. Is there an early demo available for testing? Which p...
zRzRzRzRzRzRzR closed an issue on THUDM/GLM-Edge
求个QNN/OpenVINO适配教程
### Feature request / 功能建议 求个QNN/OpenVINO适配教程 ### Motivation / 动机 端侧大模型部署 ### Your contribution / 您的贡献 端侧大模型部署
zRzRzRzRzRzRzR pushed 3 commits to main zRzRzRzRzRzRzR/transformers
- Only cast `cu_seqlens` when tracing (#35016) * Only cast `cu_seqlens` when tracing * Formatting 3480cbb
- fix variable undefined bug when return_tensors is not specified in llava processing (#34953) * fix variable undefine... 9ab8c5b
- Optimize memory usage of mllama encoder (#34930) mllama encoder memory optimization c24c79e
zRzRzRzRzRzRzR closed an issue on THUDM/GLM-4-Voice
glm4voice模型只能处理语言类的输入吗？
glm4voice模型只能处理语言类的输入吗？ 对非语言的声音，如音乐等，模型有识别能力吗？
zRzRzRzRzRzRzR closed an issue on THUDM/GLM-4-Voice
请问语音预训练数据，有没有做固定音色处理？
语音token多样性较强，哪怕表达相同语义信息，tokenize之后的token也没有规律，不知道作者有没有对预训练数据做预处理呢？
zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
ImportError: cannot import name 'ActivationCasting' from 'torchao.float8.inference' in cli_demo_quantization.py
### System Info / 系統信息 torch 2.5.1+cu121 diffusers 0.31.0 torchao 0.7.0+cpu Python 3.11.10 Windows 11 ### Information / 问题信息 - [X] The official example ...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
SFT微调在sat格式下是无法正常工作的，显存是不足的，请使用diffusers版本并使用cogvideox- factory
zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
log video OOM
### System Info / 系統信息 H100 (80GB) ### Information / 问题信息 - [X] The official example scripts / 官方的示例脚本 - [ ] My own modified scripts / 我自己修改的脚本和任务 ### Reproduction / 复现过程 I can run sat sft exa...
zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
Frame Interpolation is causing an error
### System Info / 系統信息 Windows11Pro WSL2 Ubuntu 22.04.4 LTS CUDA12 RTX4090 ### Information / 问题信息 - [X] The official example scripts / 官方的示例脚本 - [X] My own modified scripts / 我自己修改的脚本和任务 ### ...
zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
Question about the VAE upsampling
Hi, I am trying to understand the logic of the CogVideoXUpsample3D. I found that it seems like, for tensors that have an odd `t` dimension, the first frame will be treated separately for spatial on...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
What does it mean? I hope to get a specific explanation. Is it different from the video effect mentioned?
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
720 * 480 is suitable for CogVideoX1.5-5B-I2V, if it is T2V, it should be 1260 * 768.