fix: update onnxruntime version for Python 3.13 compatibility

add decoding interface

edit: update datetime codes

Update: Refine logic is now disabled

Update: Refine logic is now disabled

Update: Query-based logic is now disabled.

fix: Modify code to ensure the input dtype/shape/key is what the ONNX model expects

fix: Modify code to ensure the input dtype/shape/key is what the ONNX model expects

fix: Modify code to ensure the input dtype/shape/key is what the ONNX model expects

fix: Modify code to ensure the input dtype/shape/key is what the ONNX model expects

edit planning step in main pipeline

edit: downgrade torch version to match with torchao

edit planning step in main pipeline

edit planning step in main pipeline

fix(llm_model): align token chunking and prefix handling with engine

feat(initializer): store prefix cache as input_ids tensors

feat(inference engine): add input normalization and attention_mask support

edit process request codes

apdate ui structure

apdate ui structure

add model init at container startup

fix: make on_message_submit a generator and yield streaming tuples to match Gradio outputs

refactor: pass language textbox component to UI rendering and event binding for proper i18n support

refactor: pass language textbox component to UI rendering and event binding for proper i18n support

edit login_btn.click handler's input

edit format to match translations' placeholder

add initial ui rendering

update torch version

update import block

refactor: move LLM model initialization from global scope to function-level for lazy loading

refactor: move LLM model initialization from global scope to function-level for lazy loading

refactor: switch to lazy model/prefix initialization to improve startup and UI responsiveness

fix: configure Gradio launch with explicit server_name/port and disable SSR for stable UI rendering

Edit : switch model path

Add acelerate to requirements.txt

Edit : add model load method

Edit : add model load method

fix: update torch version

fix: update torchao version

update initializer.py

Add torchao to requirements and fix torch version to match torchao

fix: use weights_only=True in torch.load to safely load state_dict

Edit llm model filename

Edit wrong code

Refactor model initialization to use hf_hub_download cache paths

Remove manual local_dir and local path variables from config.py

Refactor load_llm: use AutoConfig and direct state_dict loading

Update initializer.py to use explicit Hub filenames

Refactor config.py to separate Hub and local paths

Refactor import paths in app.py to remove 'app.' prefix