Jose Antonio Martin H.

under ipdb only public attributes of a class can be read?

2

Hi, I have a class with __attributes When in ipdb console, those attributes doesn't show in autcomplete, indeed when you try to read one of them an Attribute error is...

[REQUEST] Add the ability to include directly Pytorch Leaning Rate schedulers

3

**Describe the solution you'd like** Add the ability to include Learning Rate schedulers directly fomr Pytorch already implemented. [https://pytorch.org/docs/stable/optim.html](https://pytorch.org/docs/stable/optim.html)

Initial Multi GPU proof of concept: ```python import torch class WrapDataParallel(torch.nn.DataParallel): def __getattr__(self, name): if name in dir(self): return super().__getattr__(name) else: return getattr(super().__getattr__("module"), name) def parallelize_algo(algo): """ Simply call parallelize_algo(algo)...

[REQUEST] Add additional info for MDPDatSetd

1

Hi, it would be very useful to have additional info dict into the MDPDataset. I think It can be implemented at two levels: 1. Additional info as "INFO" dict per...

Bug in mamba remove [package] when [package] depends on cuda

2

Remove: ─────────────────────────────────────────────────────────────────────── - __cuda 11.4 0 installed KeyError: PackageRecord(_hash=5352623804000103275, name='__cuda', version='11.4', build='0', build_number=0, channel=Channel("@"), subdir='linux-64', fn='__cuda', md5='12345678901234567890123456789012', package_type='virtual_system')

Why converting observation space to np.float32?

2

Why converting observation space to np.float? https://github.com/vwxyzjn/cleanrl/blob/329b128ea8a6afe76ce25d427c4ceba7276ad50e/cleanrl/sac_continuous_action.py#L205 This breaks compatibility with Gymnasium.

Fix handling of torch "device" association

7

- [ ] I have marked all applicable categories: + [x] exception-raising bug + [ ] RL algorithm bug + [ ] documentation request (i.e. "X is missing from the...

SAC implementation consider the reduction operator as a parameter as "min" in not always the best choice

2

the "min" operator for the two critics is not always the best choice. Consider adding the reduction method as a parameter. Also, an more o less easy hack it to...

Currently, the are two places where models are downloaded: I used: `LLMWareConfig.set_home("/data/llmware")` However the data folder The `$home$` for tools in: `/data/llmware/llmware_data/model_repo` and, even setting the `home` rest of models...

Hi with some models I experience the following error: ``` RuntimeError: Error(s) in loading state_dict for LlamaForCausalLM: size mismatch for model.layers.0.self_attn.k_proj.weight: copying a param with shape torch.Size([256, 2048]) from checkpoint,...