Vincent Moens

[Refactor] Put values, lengths and offsets of NJTs together in storage

2

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #1023 * #1022

[BugFix] Fix parsing integer batch size within export

2

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #1004

No guards for frozen dataclasses and readonly configs

3

### 🚀 The feature, motivation and pitch Currently, a guard will be put for a frozen dataclass: ``` TORCH_LOGS="+guards" python -c """from dataclasses import dataclass import torch @dataclass(frozen=True) class MyDC:...

### 🐛 Describe the bug TensorDict is a MutableMapping object, and is treated as such by torch.compile: ```python import torch from tensordict import TensorDict td = TensorDict(a=1, b=2, c=True) @torch.compile(fullgraph=True)...

MPS random number generation is slow, if not hanging forever

7

### 🐛 Describe the bug The following benchmarks are at least 3x - if not 10x slower on mps than cpu on a recent macbook pro M3 ```python from torch.utils.benchmark...