Comments (6)
I was about to post the same question, any plans to merge in newer commits from upstream? I did a manual merge but there's just way too many conflicts.
from megatron-deepspeed.
@delock @jiayulu We are planning to revise our examples including those in Megatron-DeepSpeed, but we have not finalized the plan yet. Because of that, there is no planned rebase in near future.
from megatron-deepspeed.
Thanks for clarification @conglongli
from megatron-deepspeed.
@conglongli is there any update of the plan? In particular about plan for flash attention. Does Megatron-DeepSpeed plan to support flash attention training from picking from Megatron-LM, or plan to support through DeepSpeed op builder?
from megatron-deepspeed.
@delock We had an internal discussion about your interest on fast attention (this is less related to Megatron-DeepSpeed plan), and my colleague Tunji @tjruwase would like to talk to you to learn more about this. Could you contact him next week? I guess you already have his contact. Thanks.
from megatron-deepspeed.
@delock @jiayulu We have just synced with Megatron-LM and merged 1100+ commits. Please take a look at the new main branch. We actually replaced the main branch by renaming branches, so you might have to clone the repo again because it's a bit tricky to update local repo that already cloned https://docs.github.com/en/repositories/configuring-branches-and-merges-in-your-repository/managing-branches-in-your-repository/renaming-a-branch#updating-a-local-clone-after-a-branch-name-changes. There could still exist bugs, and we very welcome bug report/bug fix contributions.
from megatron-deepspeed.
Related Issues (20)
- 2nodes, 4 gpu, tp=2,pp=2, timeout
- The link (https://the-eye.eu/public/AI/pile_neox/data/BookCorpusDataset_text_document.bin ) has expired. HOT 1
- FileNotFoundError: [Errno 2] No such file or directory: 'dataset/index-cache/xxx_doc_idx.npy' HOT 4
- Unreasonably low throughput on HGX-H100s
- Loss is increasing when fine-tuning from a Megatron-Deepspeed pretrained checkpoint.
- Problem in hf2megads_weight_converter.py
- Fine-tune llama2 with sequence parallelism HOT 3
- Bugs in GPT2 Inference Example HOT 3
- [REQUEST] Could you add a new release version tag to Megatron-Deepspeed?Thanks HOT 2
- [BUG] Problems with Mixture-of-Experts (MoE) HOT 1
- Pipeline parallelism + CPU offload?
- Assertion failure when there are more than 255 tokenized data files (assert num_datasets < 255 in blendable_dataset.py)
- AttributeError: 'Namespace' object has no attribute 'deepspeed_config_dict'. Did you mean: 'deepspeed_config'? && batch = next(self.data_iterator) HOT 1
- Expert deepcopy raises PickleError
- Call for Conversion from Huggingface to Megads with MoE
- Spurious all gather performance drop.
- Sequence Parallel is incompatible with Rotary Positional Embedding HOT 4
- 屎山代码DeepSpeed HOT 3
- about the optimizer param group
- Inquiry on Sequence Parallel Support for VocabParallelEmbedding
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from megatron-deepspeed.