-
Hi,
I noticed DeepSpeed’s training engine has already provided hybrid data and pipeline parallelism and can be further combined with model parallelism such as Megatron-LM. can you provide a 3D parall…
-
In tutorial, it says "DeepSpeed’s training engine provides hybrid data and pipeline parallelism and can be further combined with model parallelism such as Megatron-LM. ". Is there any example/tutorial…
-
# Academic papers
## MASt3R-SfM: a Fully-Integrated Solution for Unconstrained Structure-from-Motion
- [논문 링크](https://arxiv.org/pdf/2409.19152)
- 알고리즘 플로우
- 결과
- 200 장 사용시 성능…
-
### Question Validation
- [X] I have searched both the documentation and discord for an answer.
### Question
When i try to ingest a singluar node using simple search everything works fine, but when…
-
**Target:** Measure the scalability of FLUX.1 on NVIDIA Hopper architecture (both H100 & H200) using different model parallelism strategies (see [Flux.1 Performance Overview](https://github.com/xdit-p…
-
Speeding up DOM accesses would eliminate bottlenecks for a wide range of applications.
I recently discovered that the spec mandate querySelectorAll() to return elements in the [document order.](https…
-
DeepSpeed Chat use tensor parallelism via hybrid engine to generate sequence in stage3 training.
I wonder if just use zero3 inference for generation is ok? So that we don't need to transform model pa…
-
Hi, thanks for your great work!
We are attempting to deploy this framework on Volta GPUs without support of Flash-Attn. I noticed there are Llama models without RmPad that doesn't required flash-at…
-
### Question Validation
- [X] I have searched both the documentation and discord for an answer.
### Question
Hi there!
**Background:**
```python
from llama_index.core import VectorStoreInde…
-
Based on ticket request #[RITM0560743](https://uabprod.service-now.com/nav_to.do?uri=%2Fsc_req_item.do%3Fsys_id%3D34fe17a91bc95d902e00eb93604bcb58), the GROMACS software is updated on Cheaha with supp…