Add batch reasoning function

The reasoning has been converted into English.

A functioning and operational plugin version

Pluginize the SignX component

Improve the display effect of the picture

A relatively perfect version

Optimize display logic (PDF saving, good samples, good display)

Add attention keyframe extraction with heatmap visualization

Add a new eighth training session, with 2000 steps.

Start the more advanced training again. Aim for 800 steps. For now, evaluate every 100 steps.

Start the more advanced training again. Aim for 800 steps. For now, evaluate every 100 steps.

Track SMKD .h5 artifacts via LFS

Then, after the initial version training, dropout was added, and the training was continued until 38 epochs.

Add frame-annotated gloss files and documentation

Remove "Position" to make the display more tidy.

A more perfect version requires checking whether the segmentation is correct.

Increase the threshold corresponding to the feature-frame; add the corresponding original image of smkd

Add a progress bar corresponding to the gloss that can control the threshold.

Relative time and the original frame can now be displayed.

Successfully completed the attention-based segmentation inference

The reasoning script no longer deteriorates. Although sltunet requires more training to remove unknown tokens; the @ symbol has been removed.

Restore SignX files from pre-reset snapshot

Attempted to submit 4 changes, although the reasoning degraded, the reasoning could still run.