-
### Description
It is unnecessary to send the archive metadata with all transfers, we should be able to restrict it to transfers to certain destination RSEs, maybe through a new RSE attribute (archiv…
-
### 🐛 Describe the bug
After using torch.compile, I noticed a change in my output:
```
AssertionError: Tensor-likes are not close!
Mismatched elements: 8309 / 967680 (0.9%)
Greatest absolute di…
-
### 🐛 Describe the bug
When the model is compiled using torch.compile the backward graph has dynamic shaped tensors but the same model when wrapped with compiled_autograd.enable() the backward grap…
-
### 🐛 Describe the bug
By default, `torch.cdist` currently does not accept half-precision datatypes as input, throwing a `RuntimeError: "cdist" not implemented for 'Half'`.
However, if the inputs …
-
# Lines of code
# Vulnerability details
Mitigation is pretty simple here - issue was that cancel unstake doesn’t payout rewards, mitigation is adding a payout reward.
There doesn’t seem to have a…
-
In an effort to ensure [our risk identification and mitigation framework](https://github.com/finos/ai-readiness/issues/10) adheres to industry standards we intend to map our known risks and mitigation…
-
### 🐛 Describe the bug
`torch.linalg.eigh` crashes if the model is compiled into an AOTInductor model and used from the C++ side. The example python code is attached as follows:
```python
import os…
-
# Lines of code
# Vulnerability details
## Comments
in the previous implementation
in `loanLiquidation()`
Execute `IFeeManager(getFeeManager).processFees(_received, 0);` defaults `_interest==0`
B…
-
# Lines of code
# Vulnerability details
## Comments
in the previous implementation
`emitLoan()` does not limit `maxTranches`.
Users can add as many `tranches` as they want with `addNewTranche()`,…
-
# Lines of code
# Vulnerability details
## Comments
in the previous implementation
the executionData.duration time is not limited, it can lead to far exceeding the borrowing time offer.duration.
…