-
### Multi-node TPU Training with JAX
The [multi-GPU JAX training guide](https://keras.io/guides/distributed_training_with_jax/) is helpful, but it's unclear how to extend this to multi-node TPU set…
-
### Is there an existing issue for this feature request?
- [X] I have searched the existing issues
### Is your feature request related to a problem?
This pertains to the Bambu line of printers. I …
-
PGX on TPUs seems to be slower than CPUs.
With a TPU v3-8, PGX is only achieving 1638 steps / sec on the game of chess.
**Minimal Reproducible Example**
PGX CPU vs TPU Test (512 env) (with shard…
wtedw updated
2 months ago
-
### Description
i am checking jax 0.4.35's pip wheel metadata and found that `cuda12-pip` variant depends on `jaxlib==0.4.34` is this expected?
```
Metadata-Version: 2.1
Name: jax
Version: 0.4.…
-
Add profiling path to llm models on pytorch
- https://github.com/gclouduniverse/reproducibility/blob/main/Training/TPU-v5p/Llama2-7B-PyTorch/env.sh
- https://github.com/gclouduniverse/reproducibil…
-
integrate CKKS into HEIR
-
### Describe the bug
wandb version: 0.17.7
pytorch lightning version: 2.4
I am using `WandbLogger` following this [guide](https://docs.wandb.ai/guides/integrations/lightning) on TPU.
When I use …
-
# Progress
- [x] Implement TPU executor that works on a single TPU chip (without tensor parallelism) #5292
- [x] Support single-host tensor parallel inference #5871
- [x] Support multi-host ten…
-
### 🚀 The feature, motivation and pitch
I would like to serve smaller models (e.g facebook/opt-125m) using VLLM on TPU. I can't do this currently because the Pallas backend has the limitation `NotImp…
-
### Is there an existing issue for this feature request?
- [X] I have searched the existing issues
### Is your feature request related to a problem?
Update Filaments to add Bambu TPU For AMS and wo…