-
When training without providing the `mixed_precision` argument to FSDP, there is an error related to dtype mismatch in `dinov2/layers/block.py`. Is this expected?
Full stacktrace:
```txt
File "/.…
-
At the moment we call `using_spinedb` with the DB url to generate the SpineOpt data API in the global scope. This is so we can call `unit()`, `node()` and stuff like that from anywhere, which sounds n…
-
We are trying to construct models of biological phenomena, but as with any model, I'm sure there is some form of "over" and "under" fitting, but I am unsure what those things would be and how we can c…
-
SpineOpt is slow in generating large models as has been reported by many users. This is especially problematic for non-rolling models where the model generation is larger part of the total execution t…
-
I have a few questions that I hope will not much of your time.
- Is there support for IPA or some other phonetic pronunciation for words that are incorrectly pronounced or that you have a specific …
-
## 🚀 Feature
It would be quite useful to have the general ability, to compute a 'detached' `log_prob` for any distribution; i.e. blocking all gradient computation w.r.t that distribution's parameters…
-
O έλεγχος `fstisstochastic` που περιείχε το `timit_format_data.sh` αποτυγχάνει για το FST που φτιάχνουμε στο βήμα 7 του ερωτήματος 4.2. Αυτό είναι αναμενόμενο ή έχουμε κάνει λάθος σε κάτι;
Το ίδιο …
-
Things like `classes=da.unique(y)` may be inefficient. This will have to be called on each block of data, which is expensive especially if the `y` isn't persisted.
Things like `sample_weight` are t…
-
**Describe the bug**
when fine-tuning my model using deepspeed==0.13.5, and huggingface trainer, loss and grad_norm will be nan at step 2
![image](https://github.com/microsoft/DeepSpeed/assets/29994…
-
This outlines a roadmap for basic statistical functionality that Julia needs to offer. It is heavily drawn from the table of contents for MASS.
- [ ] Data processing [DataFrames.jl](https://github.com…