From c3953a8ae7473624db9e2724575d7b5df0c7eaf6 Mon Sep 17 00:00:00 2001 From: ProGamerGov Date: Sat, 4 Jun 2022 17:30:04 -0600 Subject: [PATCH 1/2] Add the CLIP Text Feature Visualization Tutorial --- ...LIP_TextFeatureVisAndSearch_OptimViz.ipynb | 1960 +++++++++++++++++ 1 file changed, 1960 insertions(+) create mode 100644 tutorials/optimviz/CLIP_TextFeatureVisAndSearch_OptimViz.ipynb diff --git a/tutorials/optimviz/CLIP_TextFeatureVisAndSearch_OptimViz.ipynb b/tutorials/optimviz/CLIP_TextFeatureVisAndSearch_OptimViz.ipynb new file mode 100644 index 0000000000..c0c6c86737 --- /dev/null +++ b/tutorials/optimviz/CLIP_TextFeatureVisAndSearch_OptimViz.ipynb @@ -0,0 +1,1960 @@ +{ + "nbformat": 4, + "nbformat_minor": 0, + "metadata": { + "colab": { + "name": "CLIP_TextFeatureVisAndSearch_OptimViz.ipynb", + "provenance": [], + "collapsed_sections": [] + }, + "kernelspec": { + "name": "python3", + "display_name": "Python 3" + }, + "language_info": { + "name": "python" + }, + "widgets": { + "application/vnd.jupyter.widget-state+json": { + "4d98f277c7b44d53b463d172ecec7d23": { + "model_module": "@jupyter-widgets/controls", + "model_name": "HBoxModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "HBoxModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "HBoxView", + "box_style": "", + "children": [ + "IPY_MODEL_b0779c6d342b47caa6e22c036e2e13e2", + "IPY_MODEL_cf6d4296a2084c598a9646b22fe08ac3", + "IPY_MODEL_4cb7cf8553de4dcbabf33c9c0798a27c" + ], + "layout": "IPY_MODEL_c2f97e1a90b644118290a860d3fc3fb2" + } + }, + "b0779c6d342b47caa6e22c036e2e13e2": { + "model_module": "@jupyter-widgets/controls", + "model_name": "HTMLModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "HTMLModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "HTMLView", + "description": "", + "description_tooltip": null, + "layout": "IPY_MODEL_dd15379bd9534719ab4488c2270b077f", + "placeholder": "​", + "style": "IPY_MODEL_7e1bbde93d924f2b93c74c694678a0fd", + "value": "100%" + } + }, + "cf6d4296a2084c598a9646b22fe08ac3": { + "model_module": "@jupyter-widgets/controls", + "model_name": "FloatProgressModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "FloatProgressModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "ProgressView", + "bar_style": "success", + "description": "", + "description_tooltip": null, + "layout": "IPY_MODEL_da6745763b764c9c9d026e316c6e76dd", + "max": 1544, + "min": 0, + "orientation": "horizontal", + "style": "IPY_MODEL_fd60edb03c1242569f3b5a17686ed2b0", + "value": 1544 + } + }, + "4cb7cf8553de4dcbabf33c9c0798a27c": { + "model_module": "@jupyter-widgets/controls", + "model_name": "HTMLModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "HTMLModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "HTMLView", + "description": "", + "description_tooltip": null, + "layout": "IPY_MODEL_693faf3b70ca489aafcb3095e04b97a3", + "placeholder": "​", + "style": "IPY_MODEL_74b0b256df6c46658082981f3d82f17a", + "value": " 1544/1544 [01:30<00:00, 17.33it/s]" + } + }, + "c2f97e1a90b644118290a860d3fc3fb2": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "dd15379bd9534719ab4488c2270b077f": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "7e1bbde93d924f2b93c74c694678a0fd": { + "model_module": "@jupyter-widgets/controls", + "model_name": "DescriptionStyleModel", + "model_module_version": "1.5.0", + "state": { + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "DescriptionStyleModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "StyleView", + "description_width": "" + } + }, + "da6745763b764c9c9d026e316c6e76dd": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "fd60edb03c1242569f3b5a17686ed2b0": { + "model_module": "@jupyter-widgets/controls", + "model_name": "ProgressStyleModel", + "model_module_version": "1.5.0", + "state": { + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "ProgressStyleModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "StyleView", + "bar_color": null, + "description_width": "" + } + }, + "693faf3b70ca489aafcb3095e04b97a3": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "74b0b256df6c46658082981f3d82f17a": { + "model_module": "@jupyter-widgets/controls", + "model_name": "DescriptionStyleModel", + "model_module_version": "1.5.0", + "state": { + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "DescriptionStyleModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "StyleView", + "description_width": "" + } + }, + "2477cf00bf934608ba560d35b5086b04": { + "model_module": "@jupyter-widgets/controls", + "model_name": "HBoxModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "HBoxModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "HBoxView", + "box_style": "", + "children": [ + "IPY_MODEL_60ffccf308304f0b9be9a7637dc3b673", + "IPY_MODEL_57aa9158b0a54edda567e76ec7ab26cc", + "IPY_MODEL_5a56829ac4724d3b848a77c43282d090" + ], + "layout": "IPY_MODEL_229423305a32404cb15dd1052b0f6f8d" + } + }, + "60ffccf308304f0b9be9a7637dc3b673": { + "model_module": "@jupyter-widgets/controls", + "model_name": "HTMLModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "HTMLModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "HTMLView", + "description": "", + "description_tooltip": null, + "layout": "IPY_MODEL_bacb750121b349d9b6276b79351c3179", + "placeholder": "​", + "style": "IPY_MODEL_8839b04f3c474d51ae3b90673a3f70bf", + "value": "100%" + } + }, + "57aa9158b0a54edda567e76ec7ab26cc": { + "model_module": "@jupyter-widgets/controls", + "model_name": "FloatProgressModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "FloatProgressModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "ProgressView", + "bar_style": "success", + "description": "", + "description_tooltip": null, + "layout": "IPY_MODEL_1b388192a3774856b551b942a6d98bd3", + "max": 1544, + "min": 0, + "orientation": "horizontal", + "style": "IPY_MODEL_c06cd0c84cd541e0a0ee19584bcaf21d", + "value": 1544 + } + }, + "5a56829ac4724d3b848a77c43282d090": { + "model_module": "@jupyter-widgets/controls", + "model_name": "HTMLModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "HTMLModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "HTMLView", + "description": "", + "description_tooltip": null, + "layout": "IPY_MODEL_b0f9363100834b32a895ea4ece0a26ec", + "placeholder": "​", + "style": "IPY_MODEL_ad418495991f4f769a643947f857aa45", + "value": " 1544/1544 [22:37<00:00, 1.11it/s]" + } + }, + "229423305a32404cb15dd1052b0f6f8d": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "bacb750121b349d9b6276b79351c3179": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "8839b04f3c474d51ae3b90673a3f70bf": { + "model_module": "@jupyter-widgets/controls", + "model_name": "DescriptionStyleModel", + "model_module_version": "1.5.0", + "state": { + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "DescriptionStyleModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "StyleView", + "description_width": "" + } + }, + "1b388192a3774856b551b942a6d98bd3": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "c06cd0c84cd541e0a0ee19584bcaf21d": { + "model_module": "@jupyter-widgets/controls", + "model_name": "ProgressStyleModel", + "model_module_version": "1.5.0", + "state": { + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "ProgressStyleModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "StyleView", + "bar_color": null, + "description_width": "" + } + }, + "b0f9363100834b32a895ea4ece0a26ec": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "ad418495991f4f769a643947f857aa45": { + "model_module": "@jupyter-widgets/controls", + "model_name": "DescriptionStyleModel", + "model_module_version": "1.5.0", + "state": { + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "DescriptionStyleModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "StyleView", + "description_width": "" + } + }, + "30f443aef4ff4653b1994ca2fdf6265f": { + "model_module": "@jupyter-widgets/controls", + "model_name": "HBoxModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "HBoxModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "HBoxView", + "box_style": "", + "children": [ + "IPY_MODEL_fd4f3d842ae54b78b1a36c1f9506ad3d", + "IPY_MODEL_c631661c871e49e38ad2fc4323ab83f6", + "IPY_MODEL_3f54e63232e244a584ba99060bb39bd2" + ], + "layout": "IPY_MODEL_58b037efabc542a588408a66af1ee332" + } + }, + "fd4f3d842ae54b78b1a36c1f9506ad3d": { + "model_module": "@jupyter-widgets/controls", + "model_name": "HTMLModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "HTMLModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "HTMLView", + "description": "", + "description_tooltip": null, + "layout": "IPY_MODEL_7e160f3270a845b3913690fa6374fbf8", + "placeholder": "​", + "style": "IPY_MODEL_c25761926edd47a28d3dbe25089c30ac", + "value": "100%" + } + }, + "c631661c871e49e38ad2fc4323ab83f6": { + "model_module": "@jupyter-widgets/controls", + "model_name": "FloatProgressModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "FloatProgressModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "ProgressView", + "bar_style": "success", + "description": "", + "description_tooltip": null, + "layout": "IPY_MODEL_9313b7418c60479a81683f4106e07900", + "max": 1544, + "min": 0, + "orientation": "horizontal", + "style": "IPY_MODEL_72f9820725eb4f3fa4bf23cda55377c2", + "value": 1544 + } + }, + "3f54e63232e244a584ba99060bb39bd2": { + "model_module": "@jupyter-widgets/controls", + "model_name": "HTMLModel", + "model_module_version": "1.5.0", + "state": { + "_dom_classes": [], + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "HTMLModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/controls", + "_view_module_version": "1.5.0", + "_view_name": "HTMLView", + "description": "", + "description_tooltip": null, + "layout": "IPY_MODEL_d63c8f2408874a29b0db10c18d4bbe0d", + "placeholder": "​", + "style": "IPY_MODEL_177efc0cc3194fbcbb7602a1d37a6d0c", + "value": " 1544/1544 [01:32<00:00, 16.72it/s]" + } + }, + "58b037efabc542a588408a66af1ee332": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "7e160f3270a845b3913690fa6374fbf8": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "c25761926edd47a28d3dbe25089c30ac": { + "model_module": "@jupyter-widgets/controls", + "model_name": "DescriptionStyleModel", + "model_module_version": "1.5.0", + "state": { + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "DescriptionStyleModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "StyleView", + "description_width": "" + } + }, + "9313b7418c60479a81683f4106e07900": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "72f9820725eb4f3fa4bf23cda55377c2": { + "model_module": "@jupyter-widgets/controls", + "model_name": "ProgressStyleModel", + "model_module_version": "1.5.0", + "state": { + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "ProgressStyleModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "StyleView", + "bar_color": null, + "description_width": "" + } + }, + "d63c8f2408874a29b0db10c18d4bbe0d": { + "model_module": "@jupyter-widgets/base", + "model_name": "LayoutModel", + "model_module_version": "1.2.0", + "state": { + "_model_module": "@jupyter-widgets/base", + "_model_module_version": "1.2.0", + "_model_name": "LayoutModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "LayoutView", + "align_content": null, + "align_items": null, + "align_self": null, + "border": null, + "bottom": null, + "display": null, + "flex": null, + "flex_flow": null, + "grid_area": null, + "grid_auto_columns": null, + "grid_auto_flow": null, + "grid_auto_rows": null, + "grid_column": null, + "grid_gap": null, + "grid_row": null, + "grid_template_areas": null, + "grid_template_columns": null, + "grid_template_rows": null, + "height": null, + "justify_content": null, + "justify_items": null, + "left": null, + "margin": null, + "max_height": null, + "max_width": null, + "min_height": null, + "min_width": null, + "object_fit": null, + "object_position": null, + "order": null, + "overflow": null, + "overflow_x": null, + "overflow_y": null, + "padding": null, + "right": null, + "top": null, + "visibility": null, + "width": null + } + }, + "177efc0cc3194fbcbb7602a1d37a6d0c": { + "model_module": "@jupyter-widgets/controls", + "model_name": "DescriptionStyleModel", + "model_module_version": "1.5.0", + "state": { + "_model_module": "@jupyter-widgets/controls", + "_model_module_version": "1.5.0", + "_model_name": "DescriptionStyleModel", + "_view_count": null, + "_view_module": "@jupyter-widgets/base", + "_view_module_version": "1.2.0", + "_view_name": "StyleView", + "description_width": "" + } + } + } + } + }, + "cells": [ + { + "cell_type": "markdown", + "source": [ + "# Text Feature Visualization & Text Search\n", + "This tutorial demonstrates how to search layer channels with text & how to perform text feature visualization on the CLIP ResNet 50x4 model as described in the [Multimodal Neurons in Artificial Neural Networks](https://distill.pub/2021/multimodal-neurons/) research paper." + ], + "metadata": { + "id": "6PyoP2q9bNGJ" + } + }, + { + "cell_type": "code", + "source": [ + "%load_ext autoreload\n", + "%autoreload 2\n", + "\n", + "\n", + "import html\n", + "from typing import Callable, List, Optional, Tuple, Union\n", + "from warnings import warn\n", + "\n", + "import captum.optim as opt\n", + "import regex as re\n", + "import torch\n", + "from captum.optim.models import clip_resnet50x4_text, clip_resnet50x4_image\n", + "from tqdm.auto import tqdm\n", + "\n", + "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")" + ], + "metadata": { + "id": "AFKTgxkmOG_U" + }, + "execution_count": null, + "outputs": [] + }, + { + "cell_type": "markdown", + "source": [ + "## Setup\n", + "\n", + "To start off, we'll define multiple helper functions and classes." + ], + "metadata": { + "id": "LWH8zkmZ7Gpn" + } + }, + { + "cell_type": "code", + "source": [ + "class PreprocessTextCLIP(torch.nn.Module):\n", + " \"\"\"\n", + " Preprocess text strings as per OpenAI's standard CLIP preprocessing / cleaning.\n", + "\n", + " See here for more information:\n", + " https://ftfy.readthedocs.io/en/latest/\n", + " https://docs.python.org/3/library/html.html#html.unescape\n", + " https://github.com/openai/CLIP/blob/main/clip/simple_tokenizer.py\n", + " \"\"\"\n", + "\n", + " __constants__ = [\"use_ftfy\"]\n", + "\n", + " def __init__(self) -> None:\n", + " super().__init__()\n", + " try:\n", + " import ftfy\n", + "\n", + " self.use_ftfy = True\n", + " except (ImportError, AssertionError):\n", + " warn(\n", + " \"Warning the ftfy library was not found, and thus heuristic unicode\"\n", + " + \" correction will not be used in the CLIPTokenizer preprocessing\"\n", + " + \" module. The library can be installed via 'pip install ftfy'\"\n", + " )\n", + " self.use_ftfy = False\n", + "\n", + " @torch.jit.ignore\n", + " def forward(self, x: List[str]) -> List[str]:\n", + " \"\"\"\n", + " Args:\n", + "\n", + " x (str or list of str): One or more strings to be cleaned.\n", + "\n", + " Returns:\n", + " x (str or list of str): A list of preprocessed / cleaned strings.\n", + " \"\"\"\n", + " assert all([isinstance(s, str) for s in x])\n", + " for i in range(len(x)):\n", + " # Heuristic unicode fixing (ex: mojibake)\n", + " if self.use_ftfy:\n", + " x[i] = ftfy.fix_text(x[i])\n", + "\n", + " # Convert named & numeric character references in HTML to unicode\n", + " x[i] = html.unescape(html.unescape(x[i]))\n", + "\n", + " # Remove duplicate whitespaces\n", + " x[i] = re.sub(r\"\\s+\", \" \", x[i].strip()).strip()\n", + "\n", + " # Only use lowercase characters\n", + " x[i] = x[i].lower()\n", + " return x\n", + "\n", + "\n", + "class CLIP_ResNet50x4(torch.nn.Module):\n", + " \"\"\"\n", + " Wrapper for combining the text and image portions of a CLIP model into the full\n", + " model.\n", + " \"\"\"\n", + "\n", + " def __init__(\n", + " self, image_model: torch.nn.Module, text_model: torch.nn.Module\n", + " ) -> None:\n", + " \"\"\"\n", + " Args:\n", + "\n", + " image_model (nn.Module): A PyTorch model instance that takes image inputs.\n", + " text_model (nn.Module): A PyTorch model instance that takes text inputs.\n", + " \"\"\"\n", + " super().__init__()\n", + " self.image_model = image_model\n", + " self.text_model = text_model\n", + "\n", + " def forward(\n", + " self, x: Union[Tuple[torch.Tensor, torch.Tensor], List[torch.Tensor]]\n", + " ) -> Tuple[torch.Tensor, torch.Tensor]:\n", + " \"\"\"\n", + " Args:\n", + "\n", + " x (tuple or list of torch.Tensor): A tuple or list of tensors, with the\n", + " format: [image_tensor, text_tensor].\n", + "\n", + " Returns:\n", + " logits_per_text (torch.Tensor): The model output.\n", + " \"\"\"\n", + " assert len(x) == 2\n", + " image, text = x\n", + " image_features = self.image_model(image)\n", + " text_features = self.text_model(text)\n", + "\n", + " image_features = image_features / image_features.norm(dim=-1, keepdim=True)\n", + " text_features = text_features / text_features.norm(dim=-1, keepdim=True)\n", + "\n", + " logit_scale = self.text_model.logit_scale.exp()\n", + "\n", + " logits_per_image = logit_scale * image_features @ text_features.t()\n", + " logits_per_text = logit_scale * text_features @ image_features.t()\n", + "\n", + " return logits_per_image, logits_per_text\n", + "\n", + "\n", + "def get_text_layer_attr(\n", + " model: torch.nn.Module, layer_target: torch.nn.Module, text_inputs: torch.Tensor\n", + ") -> torch.Tensor:\n", + " \"\"\"\n", + " Args:\n", + "\n", + " model (nn.Module): A PyTorch model instance.\n", + " layer_target (nn.Module): A target layer instance.\n", + " text_inputs (torch.Tensor): A text input to pass through the text portion of\n", + " the model.\n", + "\n", + " Returns\n", + " grad (torch.Tensor): Attributions for the target layer.\n", + " \"\"\"\n", + " grad = []\n", + " for i in range(text_inputs.shape[0]):\n", + " model_inputs = (\n", + " torch.nn.Parameter(torch.zeros(1, 3, 288, 288).to(text_inputs.device)),\n", + " text_inputs[i : i + 1].clone(),\n", + " )\n", + " attr_activations = opt.models.collect_activations(\n", + " model, [layer_target, model], model_inputs\n", + " )\n", + " target_activ = attr_activations[layer_target]\n", + " logit_activ = attr_activations[model][1]\n", + " grad_b = torch.autograd.grad(\n", + " outputs=logit_activ,\n", + " inputs=[target_activ],\n", + " grad_outputs=torch.ones_like(logit_activ),\n", + " )[0].detach()\n", + " grad.append(grad_b)\n", + " return torch.cat(grad, 0)\n", + "\n", + "\n", + "def int_token_tokenizer(\n", + " x: List[int],\n", + " context_length: int = 77,\n", + " start_token: int = 49406,\n", + " end_token: int = 49407,\n", + " padding_value: int = 0,\n", + " start_from_tokens: List[int] = [],\n", + " end_with_tokens: List[int] = [],\n", + ") -> torch.Tensor:\n", + " \"\"\"\n", + " Apply special tokens and padding to sets of tokens in integer list format.\n", + "\n", + " Args:\n", + "\n", + " context_length (int, optional): The required context length for the model.\n", + " Inputs with lengths less than context_length will be padded with\n", + " zeros.\n", + " Default: 77\n", + " start_token (str, optional): The starting token to place in front of each\n", + " text input. Set to None for no start token.\n", + " Default: \"<|startoftext|>\"\n", + " end_token (str, optional): The ending token to place at the end of each\n", + " text input. Set to None for no end token.\n", + " Default: \"<|endoftext|>\"\n", + " padding_value (int, optional): An integer value to use for padding token\n", + " sets to the desired context_length.\n", + " Default: 0\n", + " start_from_tokens (list of int, optional): Optionally add one or more\n", + " starting tokens to each input.\n", + " Default: []\n", + " end_with_tokens (list of int, optional): Optionally add one or more\n", + " ending tokens to each input.\n", + " Default: []\n", + "\n", + " Returns:\n", + " tokens (torch.Tensor): A tensors containing the token sets stacked across the\n", + " batch dimension.\n", + " \"\"\"\n", + " tokens = [\n", + " [start_token] + start_from_tokens + [t] + end_with_tokens + [end_token]\n", + " for t in x\n", + " ]\n", + " tokens = [\n", + " token_set + ([padding_value] * (context_length - len(token_set)))\n", + " for token_set in tokens\n", + " ]\n", + " return torch.as_tensor(tokens).int()" + ], + "metadata": { + "id": "uZSJVZRZOJAi" + }, + "execution_count": null, + "outputs": [] + }, + { + "cell_type": "markdown", + "source": [ + "We load both the image and text models, and then place them inside our `CLIP_ResNet50x4` wrapper class to create the full CLIP model. We also load the CLIP tokenizer, and some additional variables." + ], + "metadata": { + "id": "mAXbDI6i7cKw" + } + }, + { + "cell_type": "code", + "source": [ + "# Load the CLIP ResNet 50x4 model\n", + "clip_model_text = clip_resnet50x4_text(pretrained=True).eval().to(device)\n", + "clip_model_image = (\n", + " clip_resnet50x4_image(\n", + " pretrained=True, replace_relus_with_redirectedrelu=False, use_attnpool=True\n", + " )\n", + " .eval()\n", + " .to(device)\n", + ")\n", + "clip_model_full = CLIP_ResNet50x4(clip_model_image, clip_model_text)\n", + "\n", + "# Setup tokenizer\n", + "clip_tokenizer = opt.transforms.CLIPTokenizer(\n", + " pretrained_merges=True, preprocessing_module=PreprocessTextCLIP()\n", + ")\n", + "\n", + "# Setup tokenizer vocab range & logit scale\n", + "token_vocab_range = list(range(0, 49405)) # Standard CLIP tokens are [0-49405]\n", + "logit_scale = clip_model_text.logit_scale.exp()" + ], + "metadata": { + "id": "4bKGCAkAnS5c" + }, + "execution_count": null, + "outputs": [] + }, + { + "cell_type": "markdown", + "source": [ + "## Searching CLIP Image Layer Channels With Text\n", + "\n", + "This portion of the tutorial demonstrates how to use the text portion of the CLIP ResNet 50x4 model to search layer channels in the image portion of the model." + ], + "metadata": { + "id": "3-KNjxksTSQJ" + } + }, + { + "cell_type": "markdown", + "source": [ + "Below we show how to search target image layers for channels that relate to our text inputs!" + ], + "metadata": { + "id": "Z0sFRWGS7l7m" + } + }, + { + "cell_type": "code", + "source": [ + "text = \"kitten\" # Change to any text input or list of text inputs\n", + "text_inputs = clip_tokenizer(text).to(device)\n", + "\n", + "# Set target layer as penultimate image model layer\n", + "target = clip_model_full.image_model.layer4[5]\n", + "\n", + "# Get attributions for target layer in relation to given text inputs\n", + "layer_attr = get_text_layer_attr(clip_model_full, target, text_inputs)\n", + "\n", + "# Set the number of results to show\n", + "num_results = 5\n", + "\n", + "\n", + "for b in range(layer_attr.shape[0]):\n", + " # Sort results\n", + " channel_strengths = torch.stack(\n", + " [-torch.linalg.norm(layer_attr[b, i, :, :]) for i in range(layer_attr.shape[1])]\n", + " )\n", + " top_channels = torch.argsort(channel_strengths)[:num_results]\n", + "\n", + " # Show results\n", + " b_text = text if isinstance(text, str) else text[b]\n", + " print(\n", + " \"Top {} channels of the target layer for the text '{}' with the largest L2-norm: \\n {} \".format(\n", + " list(top_channels.size())[0], b_text, top_channels.tolist()\n", + " )\n", + " )\n", + " print(\" {}\".format(channel_strengths[top_channels].tolist()))" + ], + "metadata": { + "colab": { + "base_uri": "https://localhost:8080/" + }, + "id": "Bl1Tsk7izk7H", + "outputId": "f2805136-1733-487a-9f09-dee21d9d73b0" + }, + "execution_count": null, + "outputs": [ + { + "output_type": "stream", + "name": "stdout", + "text": [ + "Top 5 channels of the target layer for the text 'kitten' with the largest L2-norm: \n", + " [289, 1179, 607, 1543, 1124] \n", + " [-1.4196891784667969, -0.7648456692695618, -0.6109495759010315, -0.5101999044418335, -0.5019273161888123]\n" + ] + } + ] + }, + { + "cell_type": "markdown", + "source": [ + "We can see that the text input `\"kitten\"` corresponds most strongly to channel number `289` in the target layer. As the second strongest channel is significantly lower than the first, we can reasonably conclude that channel `289` is the image model's \"kitten\" channel." + ], + "metadata": { + "id": "V5B1jEBBGt4j" + } + }, + { + "cell_type": "markdown", + "source": [ + "## Similarity Search\n", + "\n", + "\n", + "CLIP models produce text & image embeddings that can be used to calculate the similarity between different images and text strings.\n", + "\n", + "Below we define a helper function for comparing embedding similarity, by searching through the model's entire vocab token range." + ], + "metadata": { + "id": "w9Cc8MolbtHB" + } + }, + { + "cell_type": "code", + "source": [ + "def embedding_token_search(\n", + " text_model: torch.nn.Module,\n", + " target_embeddings: torch.Tensor,\n", + " token_list: List[int],\n", + " batch_size: int = 32,\n", + " logit_scale: float = 100,\n", + " device: torch.device = torch.device(\"cpu\"),\n", + " start_from_tokens: List[int] = [],\n", + " end_with_tokens: List[int] = [],\n", + " tokenizer_fn: Callable[[List[int]], List[int]] = int_token_tokenizer,\n", + ") -> List[float]:\n", + " \"\"\"\n", + " Args:\n", + "\n", + " text_model (nn.Module): A PyTorch model instance.\n", + " target_embeddings (torch.Tensor): A set of normalized image or text embeddings\n", + " to find the maximal token for, with a shape of: [1, n_vals].\n", + " token_list (list of int): A list of tokens to search through.\n", + " batch_size (int, optional): The desired batch size to use.\n", + " Default: 32\n", + " device (torch.device, optional): The desired device to use.\n", + " Default: torch.device(\"cpu\")\n", + " start_from_tokens (list of int, optional): A list of one or more tokens to use\n", + " a prefix for the token search.\n", + " Default: []\n", + " end_with_tokens (list of int, optional): A list of one or more tokens to use\n", + " a suffix for the token search.\n", + " Default: []\n", + " tokenizer_fn (callable, optional): A function that takes a list of integer\n", + " token sets and applies padding & special tokens.\n", + " Default: int_token_tokenizer\n", + "\n", + " Returns:\n", + " logits_text_list (list of float): A list of values corresponding to the order\n", + " in token_list.\n", + " \"\"\"\n", + " assert target_embeddings.dim() == 2 and target_embeddings.shape[0] == 1\n", + " logits_text_list = []\n", + "\n", + " for i in tqdm(range(0, len(token_list), batch_size)):\n", + " # Prepare input tokens\n", + " token_batch = token_list[i : i + batch_size]\n", + " token_set = tokenizer_fn(\n", + " token_batch,\n", + " start_from_tokens=start_from_tokens,\n", + " end_with_tokens=end_with_tokens,\n", + " ).to(device)\n", + "\n", + " text_embeddings = text_model(token_set).detach()\n", + " text_embeddings = text_embeddings / text_embeddings.norm(dim=-1, keepdim=True)\n", + "\n", + " logits_per_text = logit_scale * text_embeddings @ target_embeddings.t()\n", + " logits_text_list += logits_per_text[:, 0].tolist()\n", + "\n", + " return logits_text_list" + ], + "metadata": { + "id": "yNW2B9GNKwq_" + }, + "execution_count": null, + "outputs": [] + }, + { + "cell_type": "markdown", + "source": [ + "### Text Similarity\n", + "\n", + "The similarity of two different text embeddings produced by the text portion of the model can easily be determined in the same way similarity between image and text embeddings is calculated." + ], + "metadata": { + "id": "MCBxsWuaK1Wm" + } + }, + { + "cell_type": "code", + "source": [ + "# Setup target embedding\n", + "text_input = \"machine learning\"\n", + "text_tokens = clip_tokenizer(text_input).to(device)\n", + "text_embeddings = clip_model_text(text_tokens).detach()\n", + "text_embeddings = text_embeddings / text_embeddings.norm(dim=-1, keepdim=True)\n", + "\n", + "# Compare target embedding with full token list\n", + "logits_text_list = embedding_token_search(\n", + " text_model=clip_model_text,\n", + " target_embeddings=text_embeddings,\n", + " token_list=token_vocab_range,\n", + " batch_size=32,\n", + " logit_scale=logit_scale,\n", + " device=device,\n", + ")\n", + "\n", + "# Sort results\n", + "num_tokens = 10\n", + "top_tokens_text = torch.argsort(torch.as_tensor(logits_text_list), descending=True)[\n", + " 0:num_tokens\n", + "]\n", + "\n", + "# Decode results\n", + "top_tokens_str = [clip_tokenizer.decode(t)[0] for t in top_tokens_text.unsqueeze(1)]\n", + "\n", + "# Display results\n", + "print(\n", + " \"Top {} most similar tokens for the input text is: \\n {} \".format(\n", + " num_tokens, top_tokens_text.tolist()\n", + " )\n", + ")\n", + "print(\"The top tokens decoded are: \\n {} \".format(top_tokens_str))" + ], + "metadata": { + "id": "8rCV0-_byeXf", + "outputId": "d3840f4e-ff78-4081-8a33-81e4ec671b16", + "colab": { + "base_uri": "https://localhost:8080/", + "height": 120, + "referenced_widgets": [ + "30f443aef4ff4653b1994ca2fdf6265f", + "fd4f3d842ae54b78b1a36c1f9506ad3d", + "c631661c871e49e38ad2fc4323ab83f6", + "3f54e63232e244a584ba99060bb39bd2", + "58b037efabc542a588408a66af1ee332", + "7e160f3270a845b3913690fa6374fbf8", + "c25761926edd47a28d3dbe25089c30ac", + "9313b7418c60479a81683f4106e07900", + "72f9820725eb4f3fa4bf23cda55377c2", + "d63c8f2408874a29b0db10c18d4bbe0d", + "177efc0cc3194fbcbb7602a1d37a6d0c" + ] + } + }, + "execution_count": null, + "outputs": [ + { + "output_type": "display_data", + "data": { + "text/plain": [ + " 0%| | 0/1544 [00:00" + ], + "image/png": "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\n" + }, + "metadata": { + "needs_background": "light" + } + } + ] + }, + { + "cell_type": "markdown", + "source": [ + "Next we search the full vocab token range with the image embeddings that we collected above." + ], + "metadata": { + "id": "vn5rHuwqsgfR" + } + }, + { + "cell_type": "code", + "source": [ + "# Collect text embedding similarities\n", + "logits_text_list = embedding_token_search(\n", + " text_model=clip_model_text,\n", + " target_embeddings=image_embedding,\n", + " token_list=token_vocab_range,\n", + " batch_size=32,\n", + " logit_scale=logit_scale,\n", + " device=device,\n", + ")\n", + "\n", + "# Sort results\n", + "num_tokens = 10\n", + "top_tokens_text = torch.argsort(torch.as_tensor(logits_text_list), descending=True)[\n", + " 0:num_tokens\n", + "]\n", + "\n", + "# Decode results\n", + "top_tokens_str = [clip_tokenizer.decode(t)[0] for t in top_tokens_text.unsqueeze(1)]\n", + "\n", + "# Display results\n", + "print(\n", + " \"Top {} most similar tokens for the input image is: \\n {} \".format(\n", + " num_tokens, top_tokens_text.tolist()\n", + " )\n", + ")\n", + "print(\"The top tokens decoded are: \\n {} \".format(top_tokens_str))" + ], + "metadata": { + "id": "Ey4YhZDxLCX-", + "colab": { + "base_uri": "https://localhost:8080/", + "height": 120, + "referenced_widgets": [ + "4d98f277c7b44d53b463d172ecec7d23", + "b0779c6d342b47caa6e22c036e2e13e2", + "cf6d4296a2084c598a9646b22fe08ac3", + "4cb7cf8553de4dcbabf33c9c0798a27c", + "c2f97e1a90b644118290a860d3fc3fb2", + "dd15379bd9534719ab4488c2270b077f", + "7e1bbde93d924f2b93c74c694678a0fd", + "da6745763b764c9c9d026e316c6e76dd", + "fd60edb03c1242569f3b5a17686ed2b0", + "693faf3b70ca489aafcb3095e04b97a3", + "74b0b256df6c46658082981f3d82f17a" + ] + }, + "outputId": "fd87c707-f8a5-46db-8cb9-9f7314414195" + }, + "execution_count": null, + "outputs": [ + { + "output_type": "display_data", + "data": { + "text/plain": [ + " 0%| | 0/1544 [00:00 Union[List[float], List[List[float]]]:\n", + " \"\"\"\n", + " Args:\n", + "\n", + " full_model (nn.Module): A PyTorch model instance.\n", + " target (nn.Module): The target layer to collect attributions from.\n", + " channel_index (int, optional): The desired channel index to collect\n", + " attributions for, in the target layer. Set to None for all channels.\n", + " token_list (list of int): A list of tokens to search through.\n", + " batch_size (int, optional): The desired batch size to use.\n", + " Default: 32\n", + " device (torch.device, optional): The desired device to use.\n", + " Default: torch.device(\"cpu\")\n", + " start_from_tokens (list of int, optional): A list of one or more tokens to use\n", + " a prefix for the token search.\n", + " Default: []\n", + " end_with_tokens (list of int, optional): A list of one or more tokens to use\n", + " a suffix for the token search.\n", + " Default: []\n", + " tokenizer_fn (callable, optional): A function that takes a list of integer\n", + " token sets and applies padding & special tokens.\n", + " Default: int_token_tokenizer\n", + "\n", + " Returns:\n", + " logits_text_list (list of float or list of list of float): A list of values\n", + " corresponding to the order in token_list.\n", + " \"\"\"\n", + " logits_text_list = []\n", + "\n", + " for i in tqdm(range(0, len(token_list), batch_size)):\n", + " # Prepare input tokens\n", + " token_batch = token_list[i : i + batch_size]\n", + " token_set = tokenizer_fn(\n", + " token_batch,\n", + " start_from_tokens=start_from_tokens,\n", + " end_with_tokens=end_with_tokens,\n", + " ).to(device)\n", + "\n", + " layer_attr = get_text_layer_attr(full_model, target, token_set)\n", + " for b in range(layer_attr.shape[0]):\n", + "\n", + " if channel_index:\n", + " channel_strengths = -torch.linalg.norm(\n", + " layer_attr[b, channel_index, ...]\n", + " )\n", + " else:\n", + " channel_strengths = torch.stack(\n", + " [\n", + " -torch.linalg.norm(layer_attr[b, c, ...])\n", + " for c in range(layer_attr.shape[1])\n", + " ]\n", + " )\n", + " logits_text_list += [channel_strengths.tolist()]\n", + "\n", + " return logits_text_list" + ], + "metadata": { + "id": "pkiKrT8B9gB2" + }, + "execution_count": null, + "outputs": [] + }, + { + "cell_type": "markdown", + "source": [ + "We can now collect attributions for the penultimate layer with a channel index of `289` for the image portion of the CLIP ResNet 50x4 model." + ], + "metadata": { + "id": "yU0Qp4sRKAPt" + } + }, + { + "cell_type": "code", + "source": [ + "# Desired target layer & channel index\n", + "target_layer = clip_model_full.image_model.layer4[5]\n", + "channel_index = 289\n", + "\n", + "\n", + "# Collect target attributions\n", + "logits_text_list = channel_token_search(\n", + " full_model=clip_model_full,\n", + " target=target_layer,\n", + " channel_index=channel_index,\n", + " token_list=token_vocab_range,\n", + " batch_size=32,\n", + " logit_scale=logit_scale,\n", + " device=device,\n", + ")" + ], + "metadata": { + "colab": { + "base_uri": "https://localhost:8080/", + "height": 49, + "referenced_widgets": [ + "2477cf00bf934608ba560d35b5086b04", + "60ffccf308304f0b9be9a7637dc3b673", + "57aa9158b0a54edda567e76ec7ab26cc", + "5a56829ac4724d3b848a77c43282d090", + "229423305a32404cb15dd1052b0f6f8d", + "bacb750121b349d9b6276b79351c3179", + "8839b04f3c474d51ae3b90673a3f70bf", + "1b388192a3774856b551b942a6d98bd3", + "c06cd0c84cd541e0a0ee19584bcaf21d", + "b0f9363100834b32a895ea4ece0a26ec", + "ad418495991f4f769a643947f857aa45" + ] + }, + "id": "Dizt021X7yBm", + "outputId": "96a585f5-6467-42d7-ed16-f4b1c7162db2" + }, + "execution_count": null, + "outputs": [ + { + "output_type": "display_data", + "data": { + "text/plain": [ + " 0%| | 0/1544 [00:00 Date: Sun, 10 Jul 2022 13:50:08 -0600 Subject: [PATCH 2/2] Move tutorial to clip directory --- .../{ => clip}/CLIP_TextFeatureVisAndSearch_OptimViz.ipynb | 0 1 file changed, 0 insertions(+), 0 deletions(-) rename tutorials/optimviz/{ => clip}/CLIP_TextFeatureVisAndSearch_OptimViz.ipynb (100%) diff --git a/tutorials/optimviz/CLIP_TextFeatureVisAndSearch_OptimViz.ipynb b/tutorials/optimviz/clip/CLIP_TextFeatureVisAndSearch_OptimViz.ipynb similarity index 100% rename from tutorials/optimviz/CLIP_TextFeatureVisAndSearch_OptimViz.ipynb rename to tutorials/optimviz/clip/CLIP_TextFeatureVisAndSearch_OptimViz.ipynb