-
**Describe the bug**
When I build and run the program using the latest commit from the main branch on WH_B0, and with the fast dispatch mode, I encounter a runtime error in the CreateDevice function.…
-
I am running the `dummy_llama.py` example provided for the QNN backend. Here is the command I am using:
```
python -m examples.qualcomm.scripts.dummy_llama2 -F -c -m SM8650 -b ./build_android
```…
-
We should start exporting transcriptions to BMM.
Notes:
* Json, so we need to "cut" json files based on the clips
* Alternatively we could just transcribe in parallel with export?
-
Hi,after that (https://github.com/mobiusml/hqq/issues/107) I want to reproduce your paper about HQQ+ but meet some problems when training:
1.Loss doesn't decrease but varies up and down…
-
### 🐛 Describe the bug
Running a simple torch.einsum is 250x slower. It executes a void gemv2T_kernel_val kernel from cuBLAS, which underutilises the GPU (uses a single SM).
This looks like a regr…
-
Platforms: rocm
This test was disabled because it is failing on main branch ([recent examples](http://torch-ci.com/failure/inductor%2Ftest_snode_runtime.py%3A%3AComputeBoundedTests%3A%3Atest_bmm)).…
-
# Description
When tracing the model with `compiler_args = ['--neuroncore-pipeline-cores', '4']` the tracing passes but traced model cannot be loaded back in due to the NEFF error (below). When traci…
-
### What happened?
When looping duties, AutoDuty is reporting the character as null, and has several errors in the logs.
Tested on Trust looping, see later comments with logs and results from lo…
-
-
17.12 11:49:01 [Server] INFO [BlueMap-Marker] Loading server plugin BlueMap-Marker v160
17.12 11:49:01 [Server] ERROR [BlueMap-Marker] Error initializing plugin 'bmm-paper-1.6.1.jar' in folder 'plugi…