-
A 1 layer hf llama 1b with 1 layer is too slow in Thunder.
Repro for instantiating the model:
```py
import torch
from transformers.models.llama import LlamaForCausalLM, LlamaConfig
LLAMA_3_2_…
t-vi updated
16 hours ago
-
I'm using clang-cl 19.1.3 to compile Windows WebKit.
In `JSRopeString::resolveToBufferSlow`, the storages of a local variable `workQueue` and an argument variable `buffer` overwraps.
https://githu…
fujii updated
2 weeks ago
-
Repro Settings
Windows 10
EP2/P/TF2
Model Phong+Envmap
Gamma settings are currently scaled incorrectly. This is also seen in the file preview.
Normal map green channel is also flipped after gam…
-
### 🐛 Describe the bug
Hi team, just reporting this problem. I can bypass it if I replace int with `math.trunc`.
repro:
```
class M(torch.nn.Module):
def forward(self, x):
ori_size…
-
Repro'ing short link: https://godbolt.org/z/6WGMxbj1e
Text of the form:
```
#include
vec: vector = (){};
```
will cause a monaco underflow
Sentry Issue: [COMPILER-EXPLORER-DW1](https://compiler-…
-
##### Description
When a swagger definition uses `allOf` to both include normal properties and inherit the properties of another model using `$ref` the generated model is missing the properties of …
-
TLDR; as of Keras 3.5 training a functional model with multiple outputs, when passing loss and y_true as dicts, is broken. Seems like tree.flatten is re-ordering entries in the y_true dict.
Keras v…
-
## Description
## Environment
I'm using this docker image: nvcr.io/nvidia/tensorrt:24.08-py3
**TensorRT Version**:
**NVIDIA GPU**: L40S
**NVIDIA Driver Version**: 535.183.01
**CUDA Version*…
-
**Describe the bug**
Having 3 schemas A, B and C, A references to B, references to C. When schemas are loaded C is read fine, B is read fine with its C dependency, then A wants to load load B, but r…
-
### What happened?
A common matcher such as:
```js
await expect(element(by.id('item.toggle'))).toHaveToggleValue(false);
```
seems to have stopped working when upgrading from RN 71 to 73. It …