-
This is a overview about all cases related to this issue, which can be refered to while I am opening PRs.
Sparse array has a better performance when it comes to a array whose values are mostly zero…
-
As i'm aware, the ground truth data KITTI provides are pretty sparse like shown below
…
-
**Feature**
I am interested to monitor the peak memory in bytes allocated in KLU. I see there is an entry in the Common structure Common.mempeak. It seems no easy to create an interface to the Comm…
-
**Describe the bug 🐞**
Calculating the matrix exponential of a CUDA sparse matrix fails with trowing the exception
ArgumentError: Attempt to use a freed reference.
or
Error: Scalar Indexing is d…
-
**Describe the bug**
I can only get the sparse model, no dense model without GPU.
Doing the gerrard-hall demo
-
The method name and the type hints for the `data_type` suggest that I can pass a `IndexType` as `data_type`. Doing so causes an exception during verification.
```python
indices = builtin.DenseArra…
-
### Bug description
After training the model with strategy `deepspeed_stage_3`, and converted the checkpoint with `convert_zero_checkpoint_to_fp32_state_dict`, I was trying to load the model with `…
-
**Describe the bug**
The Union DataType produced by UnionBuilder has non-nullable children Fields after appending nulls in the builder.
**To Reproduce**
Steps to reproduce the behavior: Try the f…
tfeda updated
2 months ago
-
I have encountered an issue when attempting to apply a zero-sized array to a `Dense` layer in Flux.jl. The operation fails due to a division by zero error, as the layer does not handle zero-sized inpu…
-
### 🐛 Describe the bug
使用hybrid_parallel训练vit-huge模型，相关配置如下
`from colossalai.amp import AMP_TYPE
BATCH_SIZE = 60
LEARNING_RATE = 3e-3
WEIGHT_DECAY = 0.3
NUM_EPOCHS = 300
WARMUP_EPOCHS = 32
…