Prajwal A

@vffuunnyy I had the same issue, had to add an extra statement to take [0] of that only if `splitter.split_text("\n".join([d.text for d in docs]))` was not empty, after which it...

According to the [source](https://github.com/Lightning-AI/lightning/blob/fd4697c62c059fc7b9946e84d91625ecb6efdbe5/src/lightning/pytorch/callbacks/model_checkpoint.py#L190), after training, you can access the best model path by `checkpoint_callback.best_model_path`. And this internal variable is updated during the training loop, so when a new trainer...

@aamster the metrics for the saved models are not saved I believe, only the model weights. So when you call `.fit`, it won't know which model to take up with...

@ryan597 what if there are checkpoints for different architectures saved? In that case, using the latest modification might not work. It should also check if the weights correspond to the...

On the cache side, I’d suggest we don’t strictly follow cache-dit, but instead design a thin cache interface that fits vLLM’s kv cache model and then implement TeaCache and DBCache/DBPrune...

Let me know when you start working on this, if possible I would like to chip in on the cache implementations :)

@SamitHuang @ZJY0516 thanks for the patience and feedback!

Pending tasks - [x] Test imports work - from vllm_omni.diffusion.models.flux2 import Flux2Pipeline - [x] Verify weight loading - Run pipeline.load_weights() with real Flux2 checkpoint - [ ] End-to-end generation test...

cc: @hsliuustc0106 @ZJY0516 Hey, could not test the original weights due to GPU constraints. Accd to: huggingface.co/blog/flux-2 (will need ~64GB) Additionally, they have provided quantized weights as well which I...

@ZJY0516 should we include support to load quantized weights for flux2 as well (ref: https://huggingface.co/blog/flux-2)