Ecosyste.ms: Timeline

Browse the timeline of events for every public repo on GitHub. Data updated hourly from GH Archive.

zRzRzRzRzRzRzR

zRzRzRzRzRzRzR pushed 23 commits to main zRzRzRzRzRzRzR/transformers
  • Fix the memory usage issue of logits in generate() (#34813) 9d16441
  • fix(DPT,Depth-Anything) `torch.export` (#34103) * Fix torch.export issue in dpt based models Signed-off-by: Phill... 8cadf76
  • Fix: take into account meta device (#34134) * Do not load for meta device * Make some minor improvements * Add... f297af5
  • Torchao weights only + prequantized compability (#34355) * weights only compability * better tests from code revi... 67890de
  • Fix hyperparameter search when optuna+deepseed (#34642) * Fix hyperparameter search when optuna+deepseed * Adding... bf42c3b
  • Fix CI by tweaking torchao tests (#34832) 3cb8676
  • Fix CI slack reporting issue (#34833) * fix * fix * fix * fix * fix --------- Co-authored-by: ydsh... 40821a2
  • VLMs: enable generation tests - last batch (#34484) * add tests for 3 more vlms * fix fuyu back * skip test 28fb02f
  • Change logging level from warning to info for `max_steps` overriding `num_train_epochs` (#34810) Update trainer.py d4e1acb
  • Add Nemotron GGUF Loading Support (#34725) * Add Nemotron GGUF Loading Support * fix the Nemotron architecture as... c57eafd
  • Improve gguf tensor processing (#34515) * add tensor processing system to separate logic for models * format refa... ae5cbf8
  • Fix ds nvme (#34444) * skip nested deepspeed.zero.Init call * make fixup * solve conflict * solve conflict ... d6a5c23
  • Fix heuristic scheduling for UAG (#34805) * fix heuristic schedule * fix style * fix format 1887159
  • Refactor StarCoder2 using modular (#34015) * Create modular_starcoder2.py * Update modular_starcoder2.py * upd... 4e90b99
  • Watermarking: fix order (#34849) fix watermarking order 6a912ff
  • Update checks for torch.distributed.tensor to require torch >= 2.5 (#34816) * Update checks for torch.distributed.te... 1867be6
  • Remove quantization related config from dequantized model (#34856) * Remove quantization related config from dequant... d9e6f30
  • Auto compile when static cache (#34247) * generate with compile * nits * simple * generate with compile ... 597efd2
  • Speculative decoding: Test the target distribution (to prevent issues like #32867) (#34553) * Update test_utils.py ... 42b36d7
  • smol improvements to support more flexible usage (#34857) * smol improvements to support more flexible usage * ruff 861758e
  • and 3 more ...

View on GitHub

zRzRzRzRzRzRzR pushed 1 commit to main MetaGLM/LawGLM

View on GitHub

zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
A girl riding a bike. This needs to be polished with a large model. Too short prompts do not work well in this version of the model. The convert demo includes the few-shot prompt words we provided,...

View on GitHub

zRzRzRzRzRzRzR pushed 2 commits to main THUDM/CogVideo
  • [Fix] fix rope temporal patch size 2fb763d
  • Merge pull request #538 from spacegoing/fix_rope_finetune_shape [Fix] fix rope temporal patch size d82922c

View on GitHub

zRzRzRzRzRzRzR closed a pull request on THUDM/CogVideo
[Fix] fix rope temporal patch size
See issue https://github.com/THUDM/CogVideo/issues/532 current finetune's implementation of `prepare_rotary_positional_embeddings` missing `patch_size_t` argument
zRzRzRzRzRzRzR created a comment on a pull request on huggingface/transformers
I would like to know if there are any improvements needed for this version, and also, I would like to know if @Cyrilvallez could guide me on how to modify modular_glm.py to make good changes. ``` ...

View on GitHub

zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
请查看 cogvideox-factory

View on GitHub

zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
how to finetune CogVideoX1.5-5B-T2V
怎样finetune CogVideoX1.5-5B-T2V 模型
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
请查看 cogvideox-factory

View on GitHub

zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
你是否是因为 pipe.to("cuda“）导致的，确认是否有没有enable to cpu

View on GitHub

zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
2B model finetune results
### System Info / 系統信息 ![微信图片_20240925104908](https://github.com/user-attachments/assets/da3e236b-357c-42ae-8a38-e7ab361b1b34) ### Information / 问题信息 - [X] The official example scripts / 官方的示例...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
SAT 1.5版本无法进行微调（现有代码），我们做了diffusers版本的，请关注 cogvideox-factory

View on GitHub

zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
KeyError: 'ops' when full fine-tuning CogVideoX1.5-5B-SAT
### System Info / 系統信息 pip list结果如下（都已满足`requirements.txt`和`sat/requirements.txt`）： ``` Package Version ------------------------ ----------- accelerate 1.1.1 ai...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
diffusers版本的lora 和sft都有提供，cogvideox factory仓库提供了两种方案

View on GitHub

zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
不是重点，但是确实影响，因为我们训练的长度是81帧

View on GitHub

zRzRzRzRzRzRzR starred Lightricks/LTX-Video
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
看起来这似乎并不是由于模型导致，而是torch的报错，你是在分布式训练吗？

View on GitHub

zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
CogvideX-Interpolation with CogVideoX1.5-5B-SAT or CogVideoX-5b-I2V
### Feature request / 功能建议 Hi, first of all thanks for the amazing model. :) I'm just qurious is there any way to use CogVideoX-5B model on CogvideoX-Interpolation method. It seems similar ...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
Which commit is your main branch?

View on GitHub

zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
1.5版本的I2v模型如何支持768*1360
/CogVideo/inference# python cli_demo.py --width 768 --height 1360会出错 1360*768正常 Traceback (most recent call last): File "/CogVideo/inference/cli_demo.py", line 183, in <module> generate_...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
不能 gradio是老的，新的没有做，通常这个相应时间已经超过gradio了，可能会掉线

View on GitHub

zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
Need to check it out @zhipuch

View on GitHub

zRzRzRzRzRzRzR closed an issue on THUDM/CogVideo
CogVideoX1.5-5B t2v takes 6 hours to generate a video
### System Info / 系統信息 # Name Version Build Channel _libgcc_mutex 0.1 main _openmp_mutex 5.1 ...
zRzRzRzRzRzRzR created a comment on an issue on THUDM/CogVideo
更新到最新的diffusers main分支

View on GitHub

zRzRzRzRzRzRzR created a review on a pull request on THUDM/CogVideo

View on GitHub

zRzRzRzRzRzRzR pushed 7 commits to main zRzRzRzRzRzRzR/diffusers
  • Fix beta and exponential sigmas + add tests (#9954) * Fix beta and exponential sigmas + add tests --------- Co... 637e230
  • Flux latents fix (#9929) * update * update * update * update * update * update --------- Co-aut... f6f7afa
  • [LoRA] enable LoRA for Mochi-1 (#9943) * feat: add lora support to Mochi-1. 805aa93
  • Improve control net block index for sd3 (#9758) * improve control net index --------- Co-authored-by: YiYi Xu ... 1235862
  • Update handle single blocks on _convert_xlabs_flux_lora_to_diffusers (#9915) * Update handle single blocks on _conve... 3139d39
  • fix controlnet module refactor (#9968) * fix e564abe
  • Fix prepare latent image ids and vae sample generators for flux (#9981) * fix * update expected slice cd6ca9d

View on GitHub

zRzRzRzRzRzRzR pushed 1 commit to glm-4-1108 zRzRzRzRzRzRzR/transformers
  • modify modular_glm failed 34e7229

View on GitHub

zRzRzRzRzRzRzR pushed 96 commits to glm-4-1108 zRzRzRzRzRzRzR/transformers
  • Update trainer for easier handling of accumulate, compile fixes, and proper reporting (#34511) * Update trainer for ... ef976a7
  • VLM: special multimodal Tokenizer (#34461) * kinda works * update * add tests * update * use special tok... 187439c
  • MPS: `isin_mps_friendly` can support 0D tensors (#34538) * apply fix * tested * make fixup 34927b0
  • Add text support to the Trainer's TensorBoard integration (#34418) * feat: add text support to TensorBoardCallback ... 48831b7
  • [i18n-HI] Translated TFLite page to Hindi (#34572) * [i18n-HI] Translated TFLite page to Hindi * [i18n-HI] Transl... a86bd6f
  • 🌐 [i18n-KO] Translated perf_train_special.md to Korean (#34590) * Translated to Ko, 1st version * updated _toctre... 1112c54
  • 🌐 [i18n-KO] Update README_ko.md (#33098) * Update README_ko.md Delete the blank paragraph in the language selecti... 0a6795a
  • fix TrainerState doc because num_input_tokens_seen is unused by defau… (#34593) fix TrainerState doc because num_inp... bfa021b
  • Fix Whisper CI (#34541) update Co-authored-by: ydshieh <[email protected]> eb81144
  • Skip DeepSpeed ZeRO Stage 3 model initialization when bnb (#34395) * Skip DeepSpeed ZeRO Stage 3 model initializatio... d0b1d8d
  • FIX: Broken repr of TorchAoConfig (#34560) FIX Broken repr of TorchAoConfig The __repr__ method references a non-... 5e1fd4e
  • Load sub-configs from composite configs (#34410) * save/load sub-configs * nit forgot these * fix copies * ... 893ad04
  • DistilBERT is ExecuTorch compatible (#34475) * DistillBERT is ExecuTorch compatible * [run_slow] distilbert * ... 663c851
  • Remove unused test_dataset (#34516) 45b0c76
  • Revert "Fix Whisper CI" (#34605) Revert "Fix Whisper CI (#34541)" This reverts commit eb811449a2389e48930c45f84c8... 74d3824
  • Fix #34494 assistant tokens when truncated (#34531) * Fix assistant tokens when truncated * fix test * fix te... 082e57e
  • Remove `@slow` for `test_eager_matches_sdpa_inference` (#34558) * update * update * update * update * up... f2d5dfb
  • Changing __repr__ in torchao to show quantized Linear (#34202) * Changing __repr__ in torchao * small update *... d2bae7e
  • Fix torchvision interpolation CI (#34539) fix-torch-interpolation-ci 9f28d0c
  • Fix `use_parallel_residual` and `qkv_bias` for StableLM GGUF config extraction (#34450) * fix stablelm qkv_bias *... e83aaaa
  • and 76 more ...

View on GitHub

zRzRzRzRzRzRzR pushed 1 commit to glm-4-1108 zRzRzRzRzRzRzR/transformers

View on GitHub

zRzRzRzRzRzRzR pushed 1 commit to glm-4-1108 zRzRzRzRzRzRzR/transformers

View on GitHub

Load more