Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / pytorch/torchtune issues and pull requests
#808 - Is there a plan to support AnswerDotAI/fsdp_qlora that fine-tune 70b LLM on 2x 24G GPUs (like: RTX 3090)?
Issue -
State: closed - Opened by yaohwang 10 months ago
- 7 comments
#806 - ImportError: cannot import name 'return_and_correct_aliasing' on Kaggle notebook
Issue -
State: closed - Opened by shure-dev 10 months ago
- 3 comments
#805 - fix llama3 tutorial link
Pull Request -
State: closed - Opened by kartikayk 10 months ago
- 3 comments
Labels: CLA Signed
#802 - Llama3-70B LoRA multi GPU
Pull Request -
State: closed - Opened by rohan-varma 10 months ago
- 4 comments
Labels: CLA Signed
#796 - fixed mixed precision in FSDP
Pull Request -
State: closed - Opened by denadai2 10 months ago
- 4 comments
#791 - Vision/Multimodal
Issue -
State: open - Opened by bhack 10 months ago
- 22 comments
Labels: enhancement
#790 - MPS support
Pull Request -
State: closed - Opened by maximegmd 10 months ago
- 22 comments
Labels: CLA Signed
#785 - Add Selective Activation Checkpointing
Pull Request -
State: closed - Opened by lessw2020 10 months ago
- 1 comment
Labels: CLA Signed
#781 - utils.set_activation_checkpointing is unnecessarily restrictive
Issue -
State: closed - Opened by rohan-varma 10 months ago
- 1 comment
#771 - DPO supports multi-device training
Pull Request -
State: closed - Opened by yechenzhi 10 months ago
- 12 comments
Labels: CLA Signed
#730 - Organise steps logic
Pull Request -
State: closed - Opened by tcapelle 10 months ago
- 9 comments
Labels: CLA Signed
#721 - Disabling LoRA with compiled models
Issue -
State: closed - Opened by BenjaminBossan 10 months ago
- 3 comments
Labels: bug
#691 - Testing mega-issue
Issue -
State: open - Opened by ebsmothers 10 months ago
Labels: enhancement
#676 - Install gcc >= 9 to support torch.compile testing with inductor backend
Issue -
State: closed - Opened by rohan-varma 11 months ago
- 1 comment
Labels: testing
#666 - Use less memory during lora state dict validation
Pull Request -
State: closed - Opened by ebsmothers 11 months ago
- 3 comments
Labels: CLA Signed
#665 - Out of CUDA on 15 GB colab . Just trying to train Mistral 7. v1
Issue -
State: closed - Opened by raymondbernard 11 months ago
- 12 comments
#654 - int4 gptq working.
Pull Request -
State: closed - Opened by HDCharles 11 months ago
- 2 comments
Labels: CLA Signed
#651 - [In Progress] FSDP2 + NF4Tensor
Pull Request -
State: closed - Opened by weifengpy 11 months ago
- 2 comments
Labels: CLA Signed
#648 - [RFC] Documenting and validating recipe params
Issue -
State: closed - Opened by RdoubleA 11 months ago
- 8 comments
#622 - Will there be a docker?
Issue -
State: closed - Opened by Playerrrrr 11 months ago
- 4 comments
#614 - fp32 Full Training seems to be taking a lot of memory
Issue -
State: closed - Opened by kartikayk 11 months ago
- 5 comments
#611 - [fix doc] AC is enabled by default
Pull Request -
State: closed - Opened by skcoirz 11 months ago
- 4 comments
Labels: CLA Signed
#609 - [RFC][fix test] missing .item() in frozen nf4 test
Pull Request -
State: closed - Opened by skcoirz 11 months ago
- 6 comments
Labels: CLA Signed
#597 - BF16 and MemoryEffient attention not working on AMD MI250
Issue -
State: closed - Opened by chauhang 11 months ago
- 5 comments
#584 - Verify that we are consistent with public/private imports
Issue -
State: closed - Opened by RdoubleA 11 months ago
Labels: wontfix
#556 - Finetuning whisper
Issue -
State: closed - Opened by AmgadHasan 11 months ago
- 1 comment
Labels: enhancement
#481 - Attach unique string to produced checkpoints
Issue -
State: closed - Opened by rohan-varma 12 months ago
Labels: wontfix
#479 - Configuring WandB logger broken / can't easily configure?
Issue -
State: closed - Opened by rohan-varma 12 months ago
- 2 comments
#454 - Separate LoRA recipe into single and multi GPU, LoRA finetune < 16GB GPU
Pull Request -
State: closed - Opened by rohan-varma 12 months ago
- 4 comments
Labels: CLA Signed
#443 - [RFC] Integration with DCP - Benchmark Results
Pull Request -
State: closed - Opened by LucasLLC 12 months ago
- 2 comments
Labels: CLA Signed
#389 - [RFC] Single Device Full Fine-tune for Llama7B in < 16GB
Pull Request -
State: open - Opened by kartikayk about 1 year ago
- 13 comments
Labels: CLA Signed
#324 - Remove `datasets` requirement and instead rely on download from `huggingface_hub`
Issue -
State: closed - Opened by joecummings about 1 year ago
- 7 comments
#301 - Move away from using `/tmp` directories
Issue -
State: closed - Opened by RdoubleA about 1 year ago
- 11 comments
#226 - Memory Profiling
Pull Request -
State: closed - Opened by msaroufim about 1 year ago
- 3 comments
Labels: CLA Signed