-
Read prompt engineering approaches specification overview for Claude and GPT:
- https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview
- https://platform.openai.com/docs/gu…
-
After talking to Kial, he told me that in previous OM2 experiments with different vertical resolutions, he noticed that the thicker the top cell, the bigger would be the wind stress sensed by the ocea…
-
```python
import torch
import thunder
from thunder.benchmarks.targets import LitGPTConfig, LitGPTBenchmark
name = "thunder"
def get_model_and_args(name):
with torch.device("cuda"):
…
-
I noticed some differences on logs that are being displayed on the Console logs view and the terminal.
The connection errors that are showing on the Terminal, are not visible on the dashboard
![…
-
- There is a difference in effective dates. This could cause implications – users submitting in BCROS, not COLIN, there is a difference in time (2 minutes) in ledger vs actual filing time (please let …
-
The Properties class is some weird custom code/inherited HashMap amalgamation. It mostly behaves the same, but during debugging crashing dacapo benchmarks, I found the following incompatibility:
In…
-
- [ ] In the current bank run example notebook... Clarify in the current example with multiple pre/post observations that this only makes sense if we assume the treatment effect to have a step-functio…
-
**Describe the bug**
When you reuse objects, the difference computation may be shortcircuited because of visitedDualValues (for cycle detection)
When you use unordered comparisions, you try in O(n^2…
-
**Describe the bug**
I was trying to figure out when to use `key` and when `key-down` events, and found these differences and peculiarities:
| key pressed | W10 `key` | W10 `key-down` | GTK `k…
-
### 🐛 Describe the bug
**Description:**
I have encountered some precision differences when using the `dropout` and `linear` in the following way.
**Code to Reproduce:**
```python
import tor…