fix: downgrade xformers to 0.0.30 for compatibility with NVIDIA 5090 …

…GPUs

fix: downgrade xformers to 0.0.30 for compatibility with NVIDIA 5090 …

…GPUs

docs: Add guidance for skipping compatibility architecture checks wit…

…h custom backend versions

docs: corrections and additions

Signed-off-by: linyinli <[email protected]>

docs: update architecture diagrams and cli options

Signed-off-by: linyinli <[email protected]>

fix: wrong allocatable from subordinate workers

ci: pin action-gh-release version

ci: pin action-gh-release version

chore: update llama-box to v0.0.153

chore: update llama-box to 0.0.144