issues
search
intelligent-machine-learning
/
glake
GLake: optimizing GPU memory management and IO transmission.
Apache License 2.0
351
stars
32
forks
source link
[Roadmap] GLake checklist
#14
Open
ruizhang1230
opened
7 months ago
ruizhang1230
commented
7 months ago
Training
[ ] Multi-stream Memory Reuse: Done, will be released
[ ] Compatible with Expandable Segment
[ ] Memory Pattern Profiling tool
[ ] DoubleOverlapping(for finetune): Done, will be released
[ ] Multipath with Specific Scenario
[ ] Compression (Lossless/Lossy for finetune)
Inference
[ ] LLM KV Cache optimization: Almost Done, will be released
[ ] MoE Inference Optmization
[ ] Other Optimization for Specific Scenario (not fragmentation)
Refactor
[ ] Support TensorFlow
[ ] Support ONNX RUNTIME
Training
Inference
Refactor