rowan-fan issues

Repositories
Issues
Comments

Results 1 issues of


                                            rowan-fan

[Usage]: Context Size Limitation and CUDA OOM with DeepSeek R1 on 2 Nodes (TP8 PP2, 16 GPUs with 141GB VRAM Each)

### Your current environment When using 2 nodes with 16 GPUs (TP8 PP2), the context size is the same as when using a single node with 8 GPUs (TP8). Setting...

usage