Initial upload: MolmoWeb-4B with HF/vLLM compatibility patches
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- .gitattributes +1027 -0
- README.md +61 -0
- added_tokens.json +305 -0
- chat_template.jinja +1 -0
- config.json +98 -0
- config.yaml +542 -0
- configuration_molmo2.py +390 -0
- generation_config.json +6 -0
- image_processing_molmo2.py +515 -0
- merges.txt +0 -0
- model-00001-of-00004.safetensors +3 -0
- model-00002-of-00004.safetensors +3 -0
- model-00003-of-00004.safetensors +3 -0
- model-00004-of-00004.safetensors +3 -0
- model.safetensors.index.json +714 -0
- modeling_molmo2.py +1764 -0
- molmoweb_logo.png +0 -0
- preprocessor_config.json +34 -0
- processing_molmo2.py +403 -0
- processor_config.json +11 -0
- special_tokens_map.json +296 -0
- tokenizer.json +0 -0
- tokenizer_config.json +2726 -0
- train/rank0.pt +3 -0
- train/rank1.pt +3 -0
- train/rank10.pt +3 -0
- train/rank11.pt +3 -0
- train/rank12.pt +3 -0
- train/rank13.pt +3 -0
- train/rank14.pt +3 -0
- train/rank15.pt +3 -0
- train/rank16.pt +3 -0
- train/rank17.pt +3 -0
- train/rank18.pt +3 -0
- train/rank19.pt +3 -0
- train/rank2.pt +3 -0
- train/rank20.pt +3 -0
- train/rank21.pt +3 -0
- train/rank22.pt +3 -0
- train/rank23.pt +3 -0
- train/rank24.pt +3 -0
- train/rank25.pt +3 -0
- train/rank26.pt +3 -0
- train/rank27.pt +3 -0
- train/rank28.pt +3 -0
- train/rank29.pt +3 -0
- train/rank3.pt +3 -0
- train/rank30.pt +3 -0
- train/rank31.pt +3 -0
- train/rank32.pt +3 -0
.gitattributes
CHANGED
|
@@ -33,3 +33,1030 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
model_and_optim/.metadata filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
model_and_optim/__0_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
model_and_optim/__0_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 39 |
+
model_and_optim/__0_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 40 |
+
model_and_optim/__0_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 41 |
+
model_and_optim/__0_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 42 |
+
model_and_optim/__0_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 43 |
+
model_and_optim/__0_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 44 |
+
model_and_optim/__0_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 45 |
+
model_and_optim/__0_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 46 |
+
model_and_optim/__0_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 47 |
+
model_and_optim/__0_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 48 |
+
model_and_optim/__0_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 49 |
+
model_and_optim/__0_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 50 |
+
model_and_optim/__0_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 51 |
+
model_and_optim/__0_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 52 |
+
model_and_optim/__0_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 53 |
+
model_and_optim/__10_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 54 |
+
model_and_optim/__10_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 55 |
+
model_and_optim/__10_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 56 |
+
model_and_optim/__10_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 57 |
+
model_and_optim/__10_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 58 |
+
model_and_optim/__10_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 59 |
+
model_and_optim/__10_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 60 |
+
model_and_optim/__10_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 61 |
+
model_and_optim/__10_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 62 |
+
model_and_optim/__10_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 63 |
+
model_and_optim/__10_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 64 |
+
model_and_optim/__10_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 65 |
+
model_and_optim/__10_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 66 |
+
model_and_optim/__10_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 67 |
+
model_and_optim/__10_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 68 |
+
model_and_optim/__10_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 69 |
+
model_and_optim/__11_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 70 |
+
model_and_optim/__11_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 71 |
+
model_and_optim/__11_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 72 |
+
model_and_optim/__11_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 73 |
+
model_and_optim/__11_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 74 |
+
model_and_optim/__11_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 75 |
+
model_and_optim/__11_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 76 |
+
model_and_optim/__11_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 77 |
+
model_and_optim/__11_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 78 |
+
model_and_optim/__11_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 79 |
+
model_and_optim/__11_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 80 |
+
model_and_optim/__11_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 81 |
+
model_and_optim/__11_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 82 |
+
model_and_optim/__11_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 83 |
+
model_and_optim/__11_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 84 |
+
model_and_optim/__11_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 85 |
+
model_and_optim/__12_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 86 |
+
model_and_optim/__12_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 87 |
+
model_and_optim/__12_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 88 |
+
model_and_optim/__12_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 89 |
+
model_and_optim/__12_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 90 |
+
model_and_optim/__12_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 91 |
+
model_and_optim/__12_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 92 |
+
model_and_optim/__12_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 93 |
+
model_and_optim/__12_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 94 |
+
model_and_optim/__12_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 95 |
+
model_and_optim/__12_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 96 |
+
model_and_optim/__12_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 97 |
+
model_and_optim/__12_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 98 |
+
model_and_optim/__12_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 99 |
+
model_and_optim/__12_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 100 |
+
model_and_optim/__12_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 101 |
+
model_and_optim/__13_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 102 |
+
model_and_optim/__13_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 103 |
+
model_and_optim/__13_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 104 |
+
model_and_optim/__13_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 105 |
+
model_and_optim/__13_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 106 |
+
model_and_optim/__13_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 107 |
+
model_and_optim/__13_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 108 |
+
model_and_optim/__13_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 109 |
+
model_and_optim/__13_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 110 |
+
model_and_optim/__13_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 111 |
+
model_and_optim/__13_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 112 |
+
model_and_optim/__13_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 113 |
+
model_and_optim/__13_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 114 |
+
model_and_optim/__13_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 115 |
+
model_and_optim/__13_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 116 |
+
model_and_optim/__13_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 117 |
+
model_and_optim/__14_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 118 |
+
model_and_optim/__14_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 119 |
+
model_and_optim/__14_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 120 |
+
model_and_optim/__14_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 121 |
+
model_and_optim/__14_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 122 |
+
model_and_optim/__14_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 123 |
+
model_and_optim/__14_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 124 |
+
model_and_optim/__14_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 125 |
+
model_and_optim/__14_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 126 |
+
model_and_optim/__14_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 127 |
+
model_and_optim/__14_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 128 |
+
model_and_optim/__14_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 129 |
+
model_and_optim/__14_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 130 |
+
model_and_optim/__14_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 131 |
+
model_and_optim/__14_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 132 |
+
model_and_optim/__14_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 133 |
+
model_and_optim/__15_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 134 |
+
model_and_optim/__15_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 135 |
+
model_and_optim/__15_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 136 |
+
model_and_optim/__15_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 137 |
+
model_and_optim/__15_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 138 |
+
model_and_optim/__15_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 139 |
+
model_and_optim/__15_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 140 |
+
model_and_optim/__15_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 141 |
+
model_and_optim/__15_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 142 |
+
model_and_optim/__15_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 143 |
+
model_and_optim/__15_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 144 |
+
model_and_optim/__15_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 145 |
+
model_and_optim/__15_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 146 |
+
model_and_optim/__15_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 147 |
+
model_and_optim/__15_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 148 |
+
model_and_optim/__15_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 149 |
+
model_and_optim/__16_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 150 |
+
model_and_optim/__16_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 151 |
+
model_and_optim/__16_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 152 |
+
model_and_optim/__16_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 153 |
+
model_and_optim/__16_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 154 |
+
model_and_optim/__16_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 155 |
+
model_and_optim/__16_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 156 |
+
model_and_optim/__16_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 157 |
+
model_and_optim/__16_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 158 |
+
model_and_optim/__16_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 159 |
+
model_and_optim/__16_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 160 |
+
model_and_optim/__16_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 161 |
+
model_and_optim/__16_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 162 |
+
model_and_optim/__16_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 163 |
+
model_and_optim/__16_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 164 |
+
model_and_optim/__16_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 165 |
+
model_and_optim/__17_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 166 |
+
model_and_optim/__17_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 167 |
+
model_and_optim/__17_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 168 |
+
model_and_optim/__17_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 169 |
+
model_and_optim/__17_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 170 |
+
model_and_optim/__17_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 171 |
+
model_and_optim/__17_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 172 |
+
model_and_optim/__17_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 173 |
+
model_and_optim/__17_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 174 |
+
model_and_optim/__17_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 175 |
+
model_and_optim/__17_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 176 |
+
model_and_optim/__17_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 177 |
+
model_and_optim/__17_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 178 |
+
model_and_optim/__17_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 179 |
+
model_and_optim/__17_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 180 |
+
model_and_optim/__17_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 181 |
+
model_and_optim/__18_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 182 |
+
model_and_optim/__18_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 183 |
+
model_and_optim/__18_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 184 |
+
model_and_optim/__18_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 185 |
+
model_and_optim/__18_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 186 |
+
model_and_optim/__18_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 187 |
+
model_and_optim/__18_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 188 |
+
model_and_optim/__18_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 189 |
+
model_and_optim/__18_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 190 |
+
model_and_optim/__18_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 191 |
+
model_and_optim/__18_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 192 |
+
model_and_optim/__18_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 193 |
+
model_and_optim/__18_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 194 |
+
model_and_optim/__18_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 195 |
+
model_and_optim/__18_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 196 |
+
model_and_optim/__18_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 197 |
+
model_and_optim/__19_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 198 |
+
model_and_optim/__19_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 199 |
+
model_and_optim/__19_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 200 |
+
model_and_optim/__19_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 201 |
+
model_and_optim/__19_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 202 |
+
model_and_optim/__19_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 203 |
+
model_and_optim/__19_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 204 |
+
model_and_optim/__19_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 205 |
+
model_and_optim/__19_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 206 |
+
model_and_optim/__19_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 207 |
+
model_and_optim/__19_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 208 |
+
model_and_optim/__19_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 209 |
+
model_and_optim/__19_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 210 |
+
model_and_optim/__19_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 211 |
+
model_and_optim/__19_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 212 |
+
model_and_optim/__19_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 213 |
+
model_and_optim/__1_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 214 |
+
model_and_optim/__1_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 215 |
+
model_and_optim/__1_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 216 |
+
model_and_optim/__1_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 217 |
+
model_and_optim/__1_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 218 |
+
model_and_optim/__1_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 219 |
+
model_and_optim/__1_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 220 |
+
model_and_optim/__1_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 221 |
+
model_and_optim/__1_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 222 |
+
model_and_optim/__1_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 223 |
+
model_and_optim/__1_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 224 |
+
model_and_optim/__1_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 225 |
+
model_and_optim/__1_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 226 |
+
model_and_optim/__1_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 227 |
+
model_and_optim/__1_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 228 |
+
model_and_optim/__1_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 229 |
+
model_and_optim/__20_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 230 |
+
model_and_optim/__20_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 231 |
+
model_and_optim/__20_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 232 |
+
model_and_optim/__20_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 233 |
+
model_and_optim/__20_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 234 |
+
model_and_optim/__20_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 235 |
+
model_and_optim/__20_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 236 |
+
model_and_optim/__20_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 237 |
+
model_and_optim/__20_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 238 |
+
model_and_optim/__20_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 239 |
+
model_and_optim/__20_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 240 |
+
model_and_optim/__20_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 241 |
+
model_and_optim/__20_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 242 |
+
model_and_optim/__20_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 243 |
+
model_and_optim/__20_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 244 |
+
model_and_optim/__20_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 245 |
+
model_and_optim/__21_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 246 |
+
model_and_optim/__21_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 247 |
+
model_and_optim/__21_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 248 |
+
model_and_optim/__21_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 249 |
+
model_and_optim/__21_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 250 |
+
model_and_optim/__21_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 251 |
+
model_and_optim/__21_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 252 |
+
model_and_optim/__21_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 253 |
+
model_and_optim/__21_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 254 |
+
model_and_optim/__21_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 255 |
+
model_and_optim/__21_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 256 |
+
model_and_optim/__21_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 257 |
+
model_and_optim/__21_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 258 |
+
model_and_optim/__21_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 259 |
+
model_and_optim/__21_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 260 |
+
model_and_optim/__21_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 261 |
+
model_and_optim/__22_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 262 |
+
model_and_optim/__22_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 263 |
+
model_and_optim/__22_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 264 |
+
model_and_optim/__22_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 265 |
+
model_and_optim/__22_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 266 |
+
model_and_optim/__22_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 267 |
+
model_and_optim/__22_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 268 |
+
model_and_optim/__22_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 269 |
+
model_and_optim/__22_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 270 |
+
model_and_optim/__22_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 271 |
+
model_and_optim/__22_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 272 |
+
model_and_optim/__22_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 273 |
+
model_and_optim/__22_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 274 |
+
model_and_optim/__22_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 275 |
+
model_and_optim/__22_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 276 |
+
model_and_optim/__22_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 277 |
+
model_and_optim/__23_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 278 |
+
model_and_optim/__23_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 279 |
+
model_and_optim/__23_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 280 |
+
model_and_optim/__23_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 281 |
+
model_and_optim/__23_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 282 |
+
model_and_optim/__23_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 283 |
+
model_and_optim/__23_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 284 |
+
model_and_optim/__23_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 285 |
+
model_and_optim/__23_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 286 |
+
model_and_optim/__23_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 287 |
+
model_and_optim/__23_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 288 |
+
model_and_optim/__23_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 289 |
+
model_and_optim/__23_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 290 |
+
model_and_optim/__23_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 291 |
+
model_and_optim/__23_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 292 |
+
model_and_optim/__23_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 293 |
+
model_and_optim/__24_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 294 |
+
model_and_optim/__24_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 295 |
+
model_and_optim/__24_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 296 |
+
model_and_optim/__24_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 297 |
+
model_and_optim/__24_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 298 |
+
model_and_optim/__24_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 299 |
+
model_and_optim/__24_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 300 |
+
model_and_optim/__24_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 301 |
+
model_and_optim/__24_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 302 |
+
model_and_optim/__24_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 303 |
+
model_and_optim/__24_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 304 |
+
model_and_optim/__24_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 305 |
+
model_and_optim/__24_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 306 |
+
model_and_optim/__24_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 307 |
+
model_and_optim/__24_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 308 |
+
model_and_optim/__24_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 309 |
+
model_and_optim/__25_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 310 |
+
model_and_optim/__25_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 311 |
+
model_and_optim/__25_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 312 |
+
model_and_optim/__25_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 313 |
+
model_and_optim/__25_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 314 |
+
model_and_optim/__25_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 315 |
+
model_and_optim/__25_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 316 |
+
model_and_optim/__25_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 317 |
+
model_and_optim/__25_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 318 |
+
model_and_optim/__25_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 319 |
+
model_and_optim/__25_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 320 |
+
model_and_optim/__25_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 321 |
+
model_and_optim/__25_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 322 |
+
model_and_optim/__25_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 323 |
+
model_and_optim/__25_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 324 |
+
model_and_optim/__25_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 325 |
+
model_and_optim/__26_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 326 |
+
model_and_optim/__26_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 327 |
+
model_and_optim/__26_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 328 |
+
model_and_optim/__26_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 329 |
+
model_and_optim/__26_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 330 |
+
model_and_optim/__26_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 331 |
+
model_and_optim/__26_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 332 |
+
model_and_optim/__26_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 333 |
+
model_and_optim/__26_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 334 |
+
model_and_optim/__26_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 335 |
+
model_and_optim/__26_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 336 |
+
model_and_optim/__26_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 337 |
+
model_and_optim/__26_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 338 |
+
model_and_optim/__26_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 339 |
+
model_and_optim/__26_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 340 |
+
model_and_optim/__26_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 341 |
+
model_and_optim/__27_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 342 |
+
model_and_optim/__27_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 343 |
+
model_and_optim/__27_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 344 |
+
model_and_optim/__27_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 345 |
+
model_and_optim/__27_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 346 |
+
model_and_optim/__27_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 347 |
+
model_and_optim/__27_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 348 |
+
model_and_optim/__27_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 349 |
+
model_and_optim/__27_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 350 |
+
model_and_optim/__27_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 351 |
+
model_and_optim/__27_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 352 |
+
model_and_optim/__27_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 353 |
+
model_and_optim/__27_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 354 |
+
model_and_optim/__27_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 355 |
+
model_and_optim/__27_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 356 |
+
model_and_optim/__27_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 357 |
+
model_and_optim/__28_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 358 |
+
model_and_optim/__28_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 359 |
+
model_and_optim/__28_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 360 |
+
model_and_optim/__28_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 361 |
+
model_and_optim/__28_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 362 |
+
model_and_optim/__28_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 363 |
+
model_and_optim/__28_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 364 |
+
model_and_optim/__28_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 365 |
+
model_and_optim/__28_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 366 |
+
model_and_optim/__28_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 367 |
+
model_and_optim/__28_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 368 |
+
model_and_optim/__28_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 369 |
+
model_and_optim/__28_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 370 |
+
model_and_optim/__28_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 371 |
+
model_and_optim/__28_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 372 |
+
model_and_optim/__28_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 373 |
+
model_and_optim/__29_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 374 |
+
model_and_optim/__29_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 375 |
+
model_and_optim/__29_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 376 |
+
model_and_optim/__29_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 377 |
+
model_and_optim/__29_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 378 |
+
model_and_optim/__29_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 379 |
+
model_and_optim/__29_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 380 |
+
model_and_optim/__29_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 381 |
+
model_and_optim/__29_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 382 |
+
model_and_optim/__29_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 383 |
+
model_and_optim/__29_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 384 |
+
model_and_optim/__29_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 385 |
+
model_and_optim/__29_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 386 |
+
model_and_optim/__29_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 387 |
+
model_and_optim/__29_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 388 |
+
model_and_optim/__29_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 389 |
+
model_and_optim/__2_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 390 |
+
model_and_optim/__2_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 391 |
+
model_and_optim/__2_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 392 |
+
model_and_optim/__2_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 393 |
+
model_and_optim/__2_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 394 |
+
model_and_optim/__2_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 395 |
+
model_and_optim/__2_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 396 |
+
model_and_optim/__2_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 397 |
+
model_and_optim/__2_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 398 |
+
model_and_optim/__2_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 399 |
+
model_and_optim/__2_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 400 |
+
model_and_optim/__2_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 401 |
+
model_and_optim/__2_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 402 |
+
model_and_optim/__2_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 403 |
+
model_and_optim/__2_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 404 |
+
model_and_optim/__2_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 405 |
+
model_and_optim/__30_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 406 |
+
model_and_optim/__30_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 407 |
+
model_and_optim/__30_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 408 |
+
model_and_optim/__30_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 409 |
+
model_and_optim/__30_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 410 |
+
model_and_optim/__30_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 411 |
+
model_and_optim/__30_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 412 |
+
model_and_optim/__30_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 413 |
+
model_and_optim/__30_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 414 |
+
model_and_optim/__30_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 415 |
+
model_and_optim/__30_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 416 |
+
model_and_optim/__30_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 417 |
+
model_and_optim/__30_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 418 |
+
model_and_optim/__30_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 419 |
+
model_and_optim/__30_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 420 |
+
model_and_optim/__30_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 421 |
+
model_and_optim/__31_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 422 |
+
model_and_optim/__31_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 423 |
+
model_and_optim/__31_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 424 |
+
model_and_optim/__31_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 425 |
+
model_and_optim/__31_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 426 |
+
model_and_optim/__31_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 427 |
+
model_and_optim/__31_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 428 |
+
model_and_optim/__31_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 429 |
+
model_and_optim/__31_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 430 |
+
model_and_optim/__31_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 431 |
+
model_and_optim/__31_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 432 |
+
model_and_optim/__31_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 433 |
+
model_and_optim/__31_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 434 |
+
model_and_optim/__31_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 435 |
+
model_and_optim/__31_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 436 |
+
model_and_optim/__31_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 437 |
+
model_and_optim/__32_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 438 |
+
model_and_optim/__32_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 439 |
+
model_and_optim/__32_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 440 |
+
model_and_optim/__32_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 441 |
+
model_and_optim/__32_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 442 |
+
model_and_optim/__32_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 443 |
+
model_and_optim/__32_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 444 |
+
model_and_optim/__32_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 445 |
+
model_and_optim/__32_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 446 |
+
model_and_optim/__32_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 447 |
+
model_and_optim/__32_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 448 |
+
model_and_optim/__32_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 449 |
+
model_and_optim/__32_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 450 |
+
model_and_optim/__32_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 451 |
+
model_and_optim/__32_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 452 |
+
model_and_optim/__32_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 453 |
+
model_and_optim/__33_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 454 |
+
model_and_optim/__33_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 455 |
+
model_and_optim/__33_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 456 |
+
model_and_optim/__33_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 457 |
+
model_and_optim/__33_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 458 |
+
model_and_optim/__33_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 459 |
+
model_and_optim/__33_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 460 |
+
model_and_optim/__33_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 461 |
+
model_and_optim/__33_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 462 |
+
model_and_optim/__33_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 463 |
+
model_and_optim/__33_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 464 |
+
model_and_optim/__33_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 465 |
+
model_and_optim/__33_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 466 |
+
model_and_optim/__33_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 467 |
+
model_and_optim/__33_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 468 |
+
model_and_optim/__33_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 469 |
+
model_and_optim/__34_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 470 |
+
model_and_optim/__34_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 471 |
+
model_and_optim/__34_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 472 |
+
model_and_optim/__34_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 473 |
+
model_and_optim/__34_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 474 |
+
model_and_optim/__34_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 475 |
+
model_and_optim/__34_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 476 |
+
model_and_optim/__34_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 477 |
+
model_and_optim/__34_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 478 |
+
model_and_optim/__34_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 479 |
+
model_and_optim/__34_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 480 |
+
model_and_optim/__34_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 481 |
+
model_and_optim/__34_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 482 |
+
model_and_optim/__34_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 483 |
+
model_and_optim/__34_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 484 |
+
model_and_optim/__34_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 485 |
+
model_and_optim/__35_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 486 |
+
model_and_optim/__35_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 487 |
+
model_and_optim/__35_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 488 |
+
model_and_optim/__35_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 489 |
+
model_and_optim/__35_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 490 |
+
model_and_optim/__35_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 491 |
+
model_and_optim/__35_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 492 |
+
model_and_optim/__35_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 493 |
+
model_and_optim/__35_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 494 |
+
model_and_optim/__35_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 495 |
+
model_and_optim/__35_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 496 |
+
model_and_optim/__35_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 497 |
+
model_and_optim/__35_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 498 |
+
model_and_optim/__35_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 499 |
+
model_and_optim/__35_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 500 |
+
model_and_optim/__35_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 501 |
+
model_and_optim/__36_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 502 |
+
model_and_optim/__36_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 503 |
+
model_and_optim/__36_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 504 |
+
model_and_optim/__36_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 505 |
+
model_and_optim/__36_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 506 |
+
model_and_optim/__36_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 507 |
+
model_and_optim/__36_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 508 |
+
model_and_optim/__36_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 509 |
+
model_and_optim/__36_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 510 |
+
model_and_optim/__36_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 511 |
+
model_and_optim/__36_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 512 |
+
model_and_optim/__36_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 513 |
+
model_and_optim/__36_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 514 |
+
model_and_optim/__36_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 515 |
+
model_and_optim/__36_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 516 |
+
model_and_optim/__36_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 517 |
+
model_and_optim/__37_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 518 |
+
model_and_optim/__37_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 519 |
+
model_and_optim/__37_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 520 |
+
model_and_optim/__37_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 521 |
+
model_and_optim/__37_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 522 |
+
model_and_optim/__37_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 523 |
+
model_and_optim/__37_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 524 |
+
model_and_optim/__37_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 525 |
+
model_and_optim/__37_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 526 |
+
model_and_optim/__37_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 527 |
+
model_and_optim/__37_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 528 |
+
model_and_optim/__37_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 529 |
+
model_and_optim/__37_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 530 |
+
model_and_optim/__37_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 531 |
+
model_and_optim/__37_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 532 |
+
model_and_optim/__37_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 533 |
+
model_and_optim/__38_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 534 |
+
model_and_optim/__38_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 535 |
+
model_and_optim/__38_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 536 |
+
model_and_optim/__38_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 537 |
+
model_and_optim/__38_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 538 |
+
model_and_optim/__38_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 539 |
+
model_and_optim/__38_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 540 |
+
model_and_optim/__38_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 541 |
+
model_and_optim/__38_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 542 |
+
model_and_optim/__38_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 543 |
+
model_and_optim/__38_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 544 |
+
model_and_optim/__38_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 545 |
+
model_and_optim/__38_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 546 |
+
model_and_optim/__38_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 547 |
+
model_and_optim/__38_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 548 |
+
model_and_optim/__38_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 549 |
+
model_and_optim/__39_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 550 |
+
model_and_optim/__39_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 551 |
+
model_and_optim/__39_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 552 |
+
model_and_optim/__39_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 553 |
+
model_and_optim/__39_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 554 |
+
model_and_optim/__39_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 555 |
+
model_and_optim/__39_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 556 |
+
model_and_optim/__39_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 557 |
+
model_and_optim/__39_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 558 |
+
model_and_optim/__39_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 559 |
+
model_and_optim/__39_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 560 |
+
model_and_optim/__39_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 561 |
+
model_and_optim/__39_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 562 |
+
model_and_optim/__39_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 563 |
+
model_and_optim/__39_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 564 |
+
model_and_optim/__39_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 565 |
+
model_and_optim/__3_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 566 |
+
model_and_optim/__3_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 567 |
+
model_and_optim/__3_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 568 |
+
model_and_optim/__3_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 569 |
+
model_and_optim/__3_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 570 |
+
model_and_optim/__3_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 571 |
+
model_and_optim/__3_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 572 |
+
model_and_optim/__3_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 573 |
+
model_and_optim/__3_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 574 |
+
model_and_optim/__3_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 575 |
+
model_and_optim/__3_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 576 |
+
model_and_optim/__3_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 577 |
+
model_and_optim/__3_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 578 |
+
model_and_optim/__3_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 579 |
+
model_and_optim/__3_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 580 |
+
model_and_optim/__3_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 581 |
+
model_and_optim/__40_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 582 |
+
model_and_optim/__40_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 583 |
+
model_and_optim/__40_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 584 |
+
model_and_optim/__40_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 585 |
+
model_and_optim/__40_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 586 |
+
model_and_optim/__40_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 587 |
+
model_and_optim/__40_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 588 |
+
model_and_optim/__40_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 589 |
+
model_and_optim/__40_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 590 |
+
model_and_optim/__40_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 591 |
+
model_and_optim/__40_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 592 |
+
model_and_optim/__40_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 593 |
+
model_and_optim/__40_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 594 |
+
model_and_optim/__40_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 595 |
+
model_and_optim/__40_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 596 |
+
model_and_optim/__40_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 597 |
+
model_and_optim/__41_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 598 |
+
model_and_optim/__41_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 599 |
+
model_and_optim/__41_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 600 |
+
model_and_optim/__41_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 601 |
+
model_and_optim/__41_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 602 |
+
model_and_optim/__41_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 603 |
+
model_and_optim/__41_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 604 |
+
model_and_optim/__41_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 605 |
+
model_and_optim/__41_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 606 |
+
model_and_optim/__41_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 607 |
+
model_and_optim/__41_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 608 |
+
model_and_optim/__41_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 609 |
+
model_and_optim/__41_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 610 |
+
model_and_optim/__41_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 611 |
+
model_and_optim/__41_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 612 |
+
model_and_optim/__41_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 613 |
+
model_and_optim/__42_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 614 |
+
model_and_optim/__42_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 615 |
+
model_and_optim/__42_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 616 |
+
model_and_optim/__42_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 617 |
+
model_and_optim/__42_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 618 |
+
model_and_optim/__42_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 619 |
+
model_and_optim/__42_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 620 |
+
model_and_optim/__42_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 621 |
+
model_and_optim/__42_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 622 |
+
model_and_optim/__42_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 623 |
+
model_and_optim/__42_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 624 |
+
model_and_optim/__42_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 625 |
+
model_and_optim/__42_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 626 |
+
model_and_optim/__42_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 627 |
+
model_and_optim/__42_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 628 |
+
model_and_optim/__42_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 629 |
+
model_and_optim/__43_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 630 |
+
model_and_optim/__43_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 631 |
+
model_and_optim/__43_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 632 |
+
model_and_optim/__43_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 633 |
+
model_and_optim/__43_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 634 |
+
model_and_optim/__43_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 635 |
+
model_and_optim/__43_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 636 |
+
model_and_optim/__43_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 637 |
+
model_and_optim/__43_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 638 |
+
model_and_optim/__43_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 639 |
+
model_and_optim/__43_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 640 |
+
model_and_optim/__43_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 641 |
+
model_and_optim/__43_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 642 |
+
model_and_optim/__43_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 643 |
+
model_and_optim/__43_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 644 |
+
model_and_optim/__43_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 645 |
+
model_and_optim/__44_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 646 |
+
model_and_optim/__44_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 647 |
+
model_and_optim/__44_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 648 |
+
model_and_optim/__44_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 649 |
+
model_and_optim/__44_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 650 |
+
model_and_optim/__44_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 651 |
+
model_and_optim/__44_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 652 |
+
model_and_optim/__44_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 653 |
+
model_and_optim/__44_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 654 |
+
model_and_optim/__44_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 655 |
+
model_and_optim/__44_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 656 |
+
model_and_optim/__44_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 657 |
+
model_and_optim/__44_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 658 |
+
model_and_optim/__44_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 659 |
+
model_and_optim/__44_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 660 |
+
model_and_optim/__44_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 661 |
+
model_and_optim/__45_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 662 |
+
model_and_optim/__45_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 663 |
+
model_and_optim/__45_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 664 |
+
model_and_optim/__45_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 665 |
+
model_and_optim/__45_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 666 |
+
model_and_optim/__45_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 667 |
+
model_and_optim/__45_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 668 |
+
model_and_optim/__45_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 669 |
+
model_and_optim/__45_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 670 |
+
model_and_optim/__45_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 671 |
+
model_and_optim/__45_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 672 |
+
model_and_optim/__45_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 673 |
+
model_and_optim/__45_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 674 |
+
model_and_optim/__45_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 675 |
+
model_and_optim/__45_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 676 |
+
model_and_optim/__45_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 677 |
+
model_and_optim/__46_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 678 |
+
model_and_optim/__46_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 679 |
+
model_and_optim/__46_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 680 |
+
model_and_optim/__46_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 681 |
+
model_and_optim/__46_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 682 |
+
model_and_optim/__46_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 683 |
+
model_and_optim/__46_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 684 |
+
model_and_optim/__46_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 685 |
+
model_and_optim/__46_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 686 |
+
model_and_optim/__46_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 687 |
+
model_and_optim/__46_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 688 |
+
model_and_optim/__46_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 689 |
+
model_and_optim/__46_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 690 |
+
model_and_optim/__46_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 691 |
+
model_and_optim/__46_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 692 |
+
model_and_optim/__46_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 693 |
+
model_and_optim/__47_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 694 |
+
model_and_optim/__47_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 695 |
+
model_and_optim/__47_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 696 |
+
model_and_optim/__47_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 697 |
+
model_and_optim/__47_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 698 |
+
model_and_optim/__47_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 699 |
+
model_and_optim/__47_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 700 |
+
model_and_optim/__47_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 701 |
+
model_and_optim/__47_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 702 |
+
model_and_optim/__47_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 703 |
+
model_and_optim/__47_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 704 |
+
model_and_optim/__47_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 705 |
+
model_and_optim/__47_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 706 |
+
model_and_optim/__47_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 707 |
+
model_and_optim/__47_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 708 |
+
model_and_optim/__47_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 709 |
+
model_and_optim/__48_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 710 |
+
model_and_optim/__48_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 711 |
+
model_and_optim/__48_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 712 |
+
model_and_optim/__48_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 713 |
+
model_and_optim/__48_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 714 |
+
model_and_optim/__48_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 715 |
+
model_and_optim/__48_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 716 |
+
model_and_optim/__48_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 717 |
+
model_and_optim/__48_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 718 |
+
model_and_optim/__48_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 719 |
+
model_and_optim/__48_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 720 |
+
model_and_optim/__48_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 721 |
+
model_and_optim/__48_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 722 |
+
model_and_optim/__48_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 723 |
+
model_and_optim/__48_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 724 |
+
model_and_optim/__48_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 725 |
+
model_and_optim/__49_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 726 |
+
model_and_optim/__49_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 727 |
+
model_and_optim/__49_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 728 |
+
model_and_optim/__49_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 729 |
+
model_and_optim/__49_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 730 |
+
model_and_optim/__49_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 731 |
+
model_and_optim/__49_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 732 |
+
model_and_optim/__49_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 733 |
+
model_and_optim/__49_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 734 |
+
model_and_optim/__49_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 735 |
+
model_and_optim/__49_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 736 |
+
model_and_optim/__49_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 737 |
+
model_and_optim/__49_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 738 |
+
model_and_optim/__49_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 739 |
+
model_and_optim/__49_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 740 |
+
model_and_optim/__49_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 741 |
+
model_and_optim/__4_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 742 |
+
model_and_optim/__4_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 743 |
+
model_and_optim/__4_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 744 |
+
model_and_optim/__4_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 745 |
+
model_and_optim/__4_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 746 |
+
model_and_optim/__4_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 747 |
+
model_and_optim/__4_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 748 |
+
model_and_optim/__4_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 749 |
+
model_and_optim/__4_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 750 |
+
model_and_optim/__4_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 751 |
+
model_and_optim/__4_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 752 |
+
model_and_optim/__4_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 753 |
+
model_and_optim/__4_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 754 |
+
model_and_optim/__4_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 755 |
+
model_and_optim/__4_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 756 |
+
model_and_optim/__4_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 757 |
+
model_and_optim/__50_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 758 |
+
model_and_optim/__50_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 759 |
+
model_and_optim/__50_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 760 |
+
model_and_optim/__50_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 761 |
+
model_and_optim/__50_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 762 |
+
model_and_optim/__50_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 763 |
+
model_and_optim/__50_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 764 |
+
model_and_optim/__50_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 765 |
+
model_and_optim/__50_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 766 |
+
model_and_optim/__50_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 767 |
+
model_and_optim/__50_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 768 |
+
model_and_optim/__50_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 769 |
+
model_and_optim/__50_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 770 |
+
model_and_optim/__50_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 771 |
+
model_and_optim/__50_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 772 |
+
model_and_optim/__50_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 773 |
+
model_and_optim/__51_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 774 |
+
model_and_optim/__51_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 775 |
+
model_and_optim/__51_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 776 |
+
model_and_optim/__51_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 777 |
+
model_and_optim/__51_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 778 |
+
model_and_optim/__51_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 779 |
+
model_and_optim/__51_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 780 |
+
model_and_optim/__51_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 781 |
+
model_and_optim/__51_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 782 |
+
model_and_optim/__51_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 783 |
+
model_and_optim/__51_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 784 |
+
model_and_optim/__51_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 785 |
+
model_and_optim/__51_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 786 |
+
model_and_optim/__51_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 787 |
+
model_and_optim/__51_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 788 |
+
model_and_optim/__51_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 789 |
+
model_and_optim/__52_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 790 |
+
model_and_optim/__52_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 791 |
+
model_and_optim/__52_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 792 |
+
model_and_optim/__52_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 793 |
+
model_and_optim/__52_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 794 |
+
model_and_optim/__52_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 795 |
+
model_and_optim/__52_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 796 |
+
model_and_optim/__52_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 797 |
+
model_and_optim/__52_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 798 |
+
model_and_optim/__52_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 799 |
+
model_and_optim/__52_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 800 |
+
model_and_optim/__52_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 801 |
+
model_and_optim/__52_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 802 |
+
model_and_optim/__52_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 803 |
+
model_and_optim/__52_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 804 |
+
model_and_optim/__52_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 805 |
+
model_and_optim/__53_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 806 |
+
model_and_optim/__53_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 807 |
+
model_and_optim/__53_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 808 |
+
model_and_optim/__53_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 809 |
+
model_and_optim/__53_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 810 |
+
model_and_optim/__53_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 811 |
+
model_and_optim/__53_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 812 |
+
model_and_optim/__53_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 813 |
+
model_and_optim/__53_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 814 |
+
model_and_optim/__53_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 815 |
+
model_and_optim/__53_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 816 |
+
model_and_optim/__53_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 817 |
+
model_and_optim/__53_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 818 |
+
model_and_optim/__53_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 819 |
+
model_and_optim/__53_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 820 |
+
model_and_optim/__53_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 821 |
+
model_and_optim/__54_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 822 |
+
model_and_optim/__54_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 823 |
+
model_and_optim/__54_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 824 |
+
model_and_optim/__54_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 825 |
+
model_and_optim/__54_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 826 |
+
model_and_optim/__54_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 827 |
+
model_and_optim/__54_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 828 |
+
model_and_optim/__54_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 829 |
+
model_and_optim/__54_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 830 |
+
model_and_optim/__54_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 831 |
+
model_and_optim/__54_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 832 |
+
model_and_optim/__54_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 833 |
+
model_and_optim/__54_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 834 |
+
model_and_optim/__54_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 835 |
+
model_and_optim/__54_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 836 |
+
model_and_optim/__54_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 837 |
+
model_and_optim/__55_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 838 |
+
model_and_optim/__55_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 839 |
+
model_and_optim/__55_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 840 |
+
model_and_optim/__55_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 841 |
+
model_and_optim/__55_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 842 |
+
model_and_optim/__55_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 843 |
+
model_and_optim/__55_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 844 |
+
model_and_optim/__55_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 845 |
+
model_and_optim/__55_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 846 |
+
model_and_optim/__55_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 847 |
+
model_and_optim/__55_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 848 |
+
model_and_optim/__55_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 849 |
+
model_and_optim/__55_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 850 |
+
model_and_optim/__55_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 851 |
+
model_and_optim/__55_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 852 |
+
model_and_optim/__55_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 853 |
+
model_and_optim/__56_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 854 |
+
model_and_optim/__56_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 855 |
+
model_and_optim/__56_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 856 |
+
model_and_optim/__56_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 857 |
+
model_and_optim/__56_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 858 |
+
model_and_optim/__56_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 859 |
+
model_and_optim/__56_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 860 |
+
model_and_optim/__56_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 861 |
+
model_and_optim/__56_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 862 |
+
model_and_optim/__56_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 863 |
+
model_and_optim/__56_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 864 |
+
model_and_optim/__56_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 865 |
+
model_and_optim/__56_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 866 |
+
model_and_optim/__56_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 867 |
+
model_and_optim/__56_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 868 |
+
model_and_optim/__56_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 869 |
+
model_and_optim/__57_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 870 |
+
model_and_optim/__57_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 871 |
+
model_and_optim/__57_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 872 |
+
model_and_optim/__57_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 873 |
+
model_and_optim/__57_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 874 |
+
model_and_optim/__57_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 875 |
+
model_and_optim/__57_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 876 |
+
model_and_optim/__57_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 877 |
+
model_and_optim/__57_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 878 |
+
model_and_optim/__57_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 879 |
+
model_and_optim/__57_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 880 |
+
model_and_optim/__57_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 881 |
+
model_and_optim/__57_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 882 |
+
model_and_optim/__57_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 883 |
+
model_and_optim/__57_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 884 |
+
model_and_optim/__57_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 885 |
+
model_and_optim/__58_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 886 |
+
model_and_optim/__58_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 887 |
+
model_and_optim/__58_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 888 |
+
model_and_optim/__58_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 889 |
+
model_and_optim/__58_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 890 |
+
model_and_optim/__58_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 891 |
+
model_and_optim/__58_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 892 |
+
model_and_optim/__58_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 893 |
+
model_and_optim/__58_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 894 |
+
model_and_optim/__58_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 895 |
+
model_and_optim/__58_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 896 |
+
model_and_optim/__58_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 897 |
+
model_and_optim/__58_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 898 |
+
model_and_optim/__58_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 899 |
+
model_and_optim/__58_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 900 |
+
model_and_optim/__58_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 901 |
+
model_and_optim/__59_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 902 |
+
model_and_optim/__59_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 903 |
+
model_and_optim/__59_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 904 |
+
model_and_optim/__59_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 905 |
+
model_and_optim/__59_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 906 |
+
model_and_optim/__59_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 907 |
+
model_and_optim/__59_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 908 |
+
model_and_optim/__59_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 909 |
+
model_and_optim/__59_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 910 |
+
model_and_optim/__59_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 911 |
+
model_and_optim/__59_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 912 |
+
model_and_optim/__59_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 913 |
+
model_and_optim/__59_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 914 |
+
model_and_optim/__59_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 915 |
+
model_and_optim/__59_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 916 |
+
model_and_optim/__59_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 917 |
+
model_and_optim/__5_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 918 |
+
model_and_optim/__5_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 919 |
+
model_and_optim/__5_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 920 |
+
model_and_optim/__5_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 921 |
+
model_and_optim/__5_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 922 |
+
model_and_optim/__5_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 923 |
+
model_and_optim/__5_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 924 |
+
model_and_optim/__5_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 925 |
+
model_and_optim/__5_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 926 |
+
model_and_optim/__5_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 927 |
+
model_and_optim/__5_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 928 |
+
model_and_optim/__5_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 929 |
+
model_and_optim/__5_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 930 |
+
model_and_optim/__5_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 931 |
+
model_and_optim/__5_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 932 |
+
model_and_optim/__5_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 933 |
+
model_and_optim/__60_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 934 |
+
model_and_optim/__60_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 935 |
+
model_and_optim/__60_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 936 |
+
model_and_optim/__60_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 937 |
+
model_and_optim/__60_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 938 |
+
model_and_optim/__60_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 939 |
+
model_and_optim/__60_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 940 |
+
model_and_optim/__60_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 941 |
+
model_and_optim/__60_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 942 |
+
model_and_optim/__60_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 943 |
+
model_and_optim/__60_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 944 |
+
model_and_optim/__60_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 945 |
+
model_and_optim/__60_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 946 |
+
model_and_optim/__60_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 947 |
+
model_and_optim/__60_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 948 |
+
model_and_optim/__60_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 949 |
+
model_and_optim/__61_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 950 |
+
model_and_optim/__61_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 951 |
+
model_and_optim/__61_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 952 |
+
model_and_optim/__61_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 953 |
+
model_and_optim/__61_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 954 |
+
model_and_optim/__61_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 955 |
+
model_and_optim/__61_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 956 |
+
model_and_optim/__61_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 957 |
+
model_and_optim/__61_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 958 |
+
model_and_optim/__61_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 959 |
+
model_and_optim/__61_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 960 |
+
model_and_optim/__61_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 961 |
+
model_and_optim/__61_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 962 |
+
model_and_optim/__61_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 963 |
+
model_and_optim/__61_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 964 |
+
model_and_optim/__61_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 965 |
+
model_and_optim/__62_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 966 |
+
model_and_optim/__62_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 967 |
+
model_and_optim/__62_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 968 |
+
model_and_optim/__62_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 969 |
+
model_and_optim/__62_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 970 |
+
model_and_optim/__62_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 971 |
+
model_and_optim/__62_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 972 |
+
model_and_optim/__62_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 973 |
+
model_and_optim/__62_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 974 |
+
model_and_optim/__62_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 975 |
+
model_and_optim/__62_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 976 |
+
model_and_optim/__62_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 977 |
+
model_and_optim/__62_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 978 |
+
model_and_optim/__62_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 979 |
+
model_and_optim/__62_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 980 |
+
model_and_optim/__62_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 981 |
+
model_and_optim/__63_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 982 |
+
model_and_optim/__63_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 983 |
+
model_and_optim/__63_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 984 |
+
model_and_optim/__63_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 985 |
+
model_and_optim/__63_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 986 |
+
model_and_optim/__63_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 987 |
+
model_and_optim/__63_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 988 |
+
model_and_optim/__63_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 989 |
+
model_and_optim/__63_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 990 |
+
model_and_optim/__63_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 991 |
+
model_and_optim/__63_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 992 |
+
model_and_optim/__63_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 993 |
+
model_and_optim/__63_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 994 |
+
model_and_optim/__63_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 995 |
+
model_and_optim/__63_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 996 |
+
model_and_optim/__63_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 997 |
+
model_and_optim/__6_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 998 |
+
model_and_optim/__6_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 999 |
+
model_and_optim/__6_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1000 |
+
model_and_optim/__6_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1001 |
+
model_and_optim/__6_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1002 |
+
model_and_optim/__6_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1003 |
+
model_and_optim/__6_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1004 |
+
model_and_optim/__6_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1005 |
+
model_and_optim/__6_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1006 |
+
model_and_optim/__6_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1007 |
+
model_and_optim/__6_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1008 |
+
model_and_optim/__6_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1009 |
+
model_and_optim/__6_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1010 |
+
model_and_optim/__6_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1011 |
+
model_and_optim/__6_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1012 |
+
model_and_optim/__6_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1013 |
+
model_and_optim/__7_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1014 |
+
model_and_optim/__7_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1015 |
+
model_and_optim/__7_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1016 |
+
model_and_optim/__7_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1017 |
+
model_and_optim/__7_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1018 |
+
model_and_optim/__7_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1019 |
+
model_and_optim/__7_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1020 |
+
model_and_optim/__7_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1021 |
+
model_and_optim/__7_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1022 |
+
model_and_optim/__7_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1023 |
+
model_and_optim/__7_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1024 |
+
model_and_optim/__7_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1025 |
+
model_and_optim/__7_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1026 |
+
model_and_optim/__7_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1027 |
+
model_and_optim/__7_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1028 |
+
model_and_optim/__7_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1029 |
+
model_and_optim/__8_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1030 |
+
model_and_optim/__8_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1031 |
+
model_and_optim/__8_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1032 |
+
model_and_optim/__8_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1033 |
+
model_and_optim/__8_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1034 |
+
model_and_optim/__8_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1035 |
+
model_and_optim/__8_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1036 |
+
model_and_optim/__8_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1037 |
+
model_and_optim/__8_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1038 |
+
model_and_optim/__8_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1039 |
+
model_and_optim/__8_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1040 |
+
model_and_optim/__8_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1041 |
+
model_and_optim/__8_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1042 |
+
model_and_optim/__8_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1043 |
+
model_and_optim/__8_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1044 |
+
model_and_optim/__8_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1045 |
+
model_and_optim/__9_0.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1046 |
+
model_and_optim/__9_1.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1047 |
+
model_and_optim/__9_10.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1048 |
+
model_and_optim/__9_11.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1049 |
+
model_and_optim/__9_12.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1050 |
+
model_and_optim/__9_13.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1051 |
+
model_and_optim/__9_14.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1052 |
+
model_and_optim/__9_15.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1053 |
+
model_and_optim/__9_2.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1054 |
+
model_and_optim/__9_3.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1055 |
+
model_and_optim/__9_4.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1056 |
+
model_and_optim/__9_5.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1057 |
+
model_and_optim/__9_6.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1058 |
+
model_and_optim/__9_7.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1059 |
+
model_and_optim/__9_8.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1060 |
+
model_and_optim/__9_9.distcp filter=lfs diff=lfs merge=lfs -text
|
| 1061 |
+
tokenizer.json filter=lfs diff=lfs merge=lfs -text
|
| 1062 |
+
molmoweb_logo.png filter=lfs diff=lfs merge=lfs -text
|
README.md
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model:
|
| 4 |
+
- allenai/MolmoWeb-4B
|
| 5 |
+
language:
|
| 6 |
+
- en
|
| 7 |
+
pipeline_tag: image-text-to-text
|
| 8 |
+
library_name: transformers
|
| 9 |
+
tags:
|
| 10 |
+
- multimodal
|
| 11 |
+
- molmo
|
| 12 |
+
- molmo2
|
| 13 |
+
- web-agent
|
| 14 |
+
- full-precision
|
| 15 |
+
- vllm-compatible
|
| 16 |
+
---
|
| 17 |
+
|
| 18 |
+
# ravilution/MolmoWeb-4B
|
| 19 |
+
|
| 20 |
+
This is a **full-precision** Hugging Face– and **vLLM-compatible** release of [`allenai/MolmoWeb-4B`](https://huggingface.co/allenai/MolmoWeb-4B), a vision-based web agent model by Ai2 capable of navigating and interacting with web browsers.
|
| 21 |
+
|
| 22 |
+
It follows the same idea as [`ravilution/MolmoWeb-8B-8bit-mlx`](https://huggingface.co/ravilution/MolmoWeb-8B-8bit-mlx): a personal Hub copy with a clear description and practical loading notes—here for the **4B** dense checkpoint rather than an MLX quantization.
|
| 23 |
+
|
| 24 |
+
> **Note:** This is a **4B** parameter model (four `safetensors` shards). A few **post-download patches** were applied locally so tokenization and generation metadata match what downstream stacks (including vLLM) expect: `eos_token_id` / `bos_token_id` / `pad_token_id`, `transformers_version` in `config.json` and `generation_config.json`, and the tokenizer pretokenizer regex (Mistral-style `(?i:...)` fix). Patches are **idempotent** if you re-run them on a fresh download.
|
| 25 |
+
|
| 26 |
+
Refer to the [original model card](https://huggingface.co/allenai/MolmoWeb-4B) for benchmarks, architecture, training data, and intended use.
|
| 27 |
+
|
| 28 |
+
## Use with Transformers
|
| 29 |
+
|
| 30 |
+
```bash
|
| 31 |
+
pip install -U transformers accelerate torch pillow
|
| 32 |
+
```
|
| 33 |
+
|
| 34 |
+
```python
|
| 35 |
+
from transformers import AutoProcessor, AutoModelForImageTextToText
|
| 36 |
+
import torch
|
| 37 |
+
|
| 38 |
+
model_id = "ravilution/MolmoWeb-4B"
|
| 39 |
+
|
| 40 |
+
model = AutoModelForImageTextToText.from_pretrained(
|
| 41 |
+
model_id,
|
| 42 |
+
trust_remote_code=True,
|
| 43 |
+
torch_dtype=torch.float32,
|
| 44 |
+
attn_implementation="sdpa",
|
| 45 |
+
device_map="auto",
|
| 46 |
+
)
|
| 47 |
+
processor = AutoProcessor.from_pretrained(
|
| 48 |
+
model_id,
|
| 49 |
+
trust_remote_code=True,
|
| 50 |
+
padding_side="left",
|
| 51 |
+
)
|
| 52 |
+
```
|
| 53 |
+
|
| 54 |
+
## Provenance
|
| 55 |
+
|
| 56 |
+
- **Upstream weights:** `allenai/MolmoWeb-4B`
|
| 57 |
+
- **Changes on top:** compatibility patches only (config / generation_config / tokenizer metadata as above); **no** retraining or architectural edits.
|
| 58 |
+
|
| 59 |
+
## License
|
| 60 |
+
|
| 61 |
+
Apache 2.0 — see the [original model](https://huggingface.co/allenai/MolmoWeb-4B) for details. Please review Ai2’s Responsible Use Guidelines for intended use.
|
added_tokens.json
ADDED
|
@@ -0,0 +1,305 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"</think>": 151668,
|
| 3 |
+
"</tool_call>": 151658,
|
| 4 |
+
"</tool_response>": 151666,
|
| 5 |
+
"<frame_end>": 151944,
|
| 6 |
+
"<frame_start>": 151943,
|
| 7 |
+
"<im_col>": 151939,
|
| 8 |
+
"<im_end>": 151937,
|
| 9 |
+
"<im_low>": 151942,
|
| 10 |
+
"<im_patch>": 151938,
|
| 11 |
+
"<im_start>": 151936,
|
| 12 |
+
"<low_res_im_start>": 151940,
|
| 13 |
+
"<think>": 151667,
|
| 14 |
+
"<tool_call>": 151657,
|
| 15 |
+
"<tool_response>": 151665,
|
| 16 |
+
"<|box_end|>": 151649,
|
| 17 |
+
"<|box_start|>": 151648,
|
| 18 |
+
"<|endoftext|>": 151643,
|
| 19 |
+
"<|file_sep|>": 151664,
|
| 20 |
+
"<|fim_middle|>": 151660,
|
| 21 |
+
"<|fim_pad|>": 151662,
|
| 22 |
+
"<|fim_prefix|>": 151659,
|
| 23 |
+
"<|fim_suffix|>": 151661,
|
| 24 |
+
"<|im_end|>": 151645,
|
| 25 |
+
"<|im_start|>": 151644,
|
| 26 |
+
"<|image_pad|>": 151655,
|
| 27 |
+
"<|image|>": 151941,
|
| 28 |
+
"<|object_ref_end|>": 151647,
|
| 29 |
+
"<|object_ref_start|>": 151646,
|
| 30 |
+
"<|quad_end|>": 151651,
|
| 31 |
+
"<|quad_start|>": 151650,
|
| 32 |
+
"<|repo_name|>": 151663,
|
| 33 |
+
"<|video_pad|>": 151656,
|
| 34 |
+
"<|video|>": 151945,
|
| 35 |
+
"<|vision_end|>": 151653,
|
| 36 |
+
"<|vision_pad|>": 151654,
|
| 37 |
+
"<|vision_start|>": 151652,
|
| 38 |
+
"|<EXTRA_TOKENS_0>|": 151669,
|
| 39 |
+
"|<EXTRA_TOKENS_100>|": 151769,
|
| 40 |
+
"|<EXTRA_TOKENS_101>|": 151770,
|
| 41 |
+
"|<EXTRA_TOKENS_102>|": 151771,
|
| 42 |
+
"|<EXTRA_TOKENS_103>|": 151772,
|
| 43 |
+
"|<EXTRA_TOKENS_104>|": 151773,
|
| 44 |
+
"|<EXTRA_TOKENS_105>|": 151774,
|
| 45 |
+
"|<EXTRA_TOKENS_106>|": 151775,
|
| 46 |
+
"|<EXTRA_TOKENS_107>|": 151776,
|
| 47 |
+
"|<EXTRA_TOKENS_108>|": 151777,
|
| 48 |
+
"|<EXTRA_TOKENS_109>|": 151778,
|
| 49 |
+
"|<EXTRA_TOKENS_10>|": 151679,
|
| 50 |
+
"|<EXTRA_TOKENS_110>|": 151779,
|
| 51 |
+
"|<EXTRA_TOKENS_111>|": 151780,
|
| 52 |
+
"|<EXTRA_TOKENS_112>|": 151781,
|
| 53 |
+
"|<EXTRA_TOKENS_113>|": 151782,
|
| 54 |
+
"|<EXTRA_TOKENS_114>|": 151783,
|
| 55 |
+
"|<EXTRA_TOKENS_115>|": 151784,
|
| 56 |
+
"|<EXTRA_TOKENS_116>|": 151785,
|
| 57 |
+
"|<EXTRA_TOKENS_117>|": 151786,
|
| 58 |
+
"|<EXTRA_TOKENS_118>|": 151787,
|
| 59 |
+
"|<EXTRA_TOKENS_119>|": 151788,
|
| 60 |
+
"|<EXTRA_TOKENS_11>|": 151680,
|
| 61 |
+
"|<EXTRA_TOKENS_120>|": 151789,
|
| 62 |
+
"|<EXTRA_TOKENS_121>|": 151790,
|
| 63 |
+
"|<EXTRA_TOKENS_122>|": 151791,
|
| 64 |
+
"|<EXTRA_TOKENS_123>|": 151792,
|
| 65 |
+
"|<EXTRA_TOKENS_124>|": 151793,
|
| 66 |
+
"|<EXTRA_TOKENS_125>|": 151794,
|
| 67 |
+
"|<EXTRA_TOKENS_126>|": 151795,
|
| 68 |
+
"|<EXTRA_TOKENS_127>|": 151796,
|
| 69 |
+
"|<EXTRA_TOKENS_128>|": 151797,
|
| 70 |
+
"|<EXTRA_TOKENS_129>|": 151798,
|
| 71 |
+
"|<EXTRA_TOKENS_12>|": 151681,
|
| 72 |
+
"|<EXTRA_TOKENS_130>|": 151799,
|
| 73 |
+
"|<EXTRA_TOKENS_131>|": 151800,
|
| 74 |
+
"|<EXTRA_TOKENS_132>|": 151801,
|
| 75 |
+
"|<EXTRA_TOKENS_133>|": 151802,
|
| 76 |
+
"|<EXTRA_TOKENS_134>|": 151803,
|
| 77 |
+
"|<EXTRA_TOKENS_135>|": 151804,
|
| 78 |
+
"|<EXTRA_TOKENS_136>|": 151805,
|
| 79 |
+
"|<EXTRA_TOKENS_137>|": 151806,
|
| 80 |
+
"|<EXTRA_TOKENS_138>|": 151807,
|
| 81 |
+
"|<EXTRA_TOKENS_139>|": 151808,
|
| 82 |
+
"|<EXTRA_TOKENS_13>|": 151682,
|
| 83 |
+
"|<EXTRA_TOKENS_140>|": 151809,
|
| 84 |
+
"|<EXTRA_TOKENS_141>|": 151810,
|
| 85 |
+
"|<EXTRA_TOKENS_142>|": 151811,
|
| 86 |
+
"|<EXTRA_TOKENS_143>|": 151812,
|
| 87 |
+
"|<EXTRA_TOKENS_144>|": 151813,
|
| 88 |
+
"|<EXTRA_TOKENS_145>|": 151814,
|
| 89 |
+
"|<EXTRA_TOKENS_146>|": 151815,
|
| 90 |
+
"|<EXTRA_TOKENS_147>|": 151816,
|
| 91 |
+
"|<EXTRA_TOKENS_148>|": 151817,
|
| 92 |
+
"|<EXTRA_TOKENS_149>|": 151818,
|
| 93 |
+
"|<EXTRA_TOKENS_14>|": 151683,
|
| 94 |
+
"|<EXTRA_TOKENS_150>|": 151819,
|
| 95 |
+
"|<EXTRA_TOKENS_151>|": 151820,
|
| 96 |
+
"|<EXTRA_TOKENS_152>|": 151821,
|
| 97 |
+
"|<EXTRA_TOKENS_153>|": 151822,
|
| 98 |
+
"|<EXTRA_TOKENS_154>|": 151823,
|
| 99 |
+
"|<EXTRA_TOKENS_155>|": 151824,
|
| 100 |
+
"|<EXTRA_TOKENS_156>|": 151825,
|
| 101 |
+
"|<EXTRA_TOKENS_157>|": 151826,
|
| 102 |
+
"|<EXTRA_TOKENS_158>|": 151827,
|
| 103 |
+
"|<EXTRA_TOKENS_159>|": 151828,
|
| 104 |
+
"|<EXTRA_TOKENS_15>|": 151684,
|
| 105 |
+
"|<EXTRA_TOKENS_160>|": 151829,
|
| 106 |
+
"|<EXTRA_TOKENS_161>|": 151830,
|
| 107 |
+
"|<EXTRA_TOKENS_162>|": 151831,
|
| 108 |
+
"|<EXTRA_TOKENS_163>|": 151832,
|
| 109 |
+
"|<EXTRA_TOKENS_164>|": 151833,
|
| 110 |
+
"|<EXTRA_TOKENS_165>|": 151834,
|
| 111 |
+
"|<EXTRA_TOKENS_166>|": 151835,
|
| 112 |
+
"|<EXTRA_TOKENS_167>|": 151836,
|
| 113 |
+
"|<EXTRA_TOKENS_168>|": 151837,
|
| 114 |
+
"|<EXTRA_TOKENS_169>|": 151838,
|
| 115 |
+
"|<EXTRA_TOKENS_16>|": 151685,
|
| 116 |
+
"|<EXTRA_TOKENS_170>|": 151839,
|
| 117 |
+
"|<EXTRA_TOKENS_171>|": 151840,
|
| 118 |
+
"|<EXTRA_TOKENS_172>|": 151841,
|
| 119 |
+
"|<EXTRA_TOKENS_173>|": 151842,
|
| 120 |
+
"|<EXTRA_TOKENS_174>|": 151843,
|
| 121 |
+
"|<EXTRA_TOKENS_175>|": 151844,
|
| 122 |
+
"|<EXTRA_TOKENS_176>|": 151845,
|
| 123 |
+
"|<EXTRA_TOKENS_177>|": 151846,
|
| 124 |
+
"|<EXTRA_TOKENS_178>|": 151847,
|
| 125 |
+
"|<EXTRA_TOKENS_179>|": 151848,
|
| 126 |
+
"|<EXTRA_TOKENS_17>|": 151686,
|
| 127 |
+
"|<EXTRA_TOKENS_180>|": 151849,
|
| 128 |
+
"|<EXTRA_TOKENS_181>|": 151850,
|
| 129 |
+
"|<EXTRA_TOKENS_182>|": 151851,
|
| 130 |
+
"|<EXTRA_TOKENS_183>|": 151852,
|
| 131 |
+
"|<EXTRA_TOKENS_184>|": 151853,
|
| 132 |
+
"|<EXTRA_TOKENS_185>|": 151854,
|
| 133 |
+
"|<EXTRA_TOKENS_186>|": 151855,
|
| 134 |
+
"|<EXTRA_TOKENS_187>|": 151856,
|
| 135 |
+
"|<EXTRA_TOKENS_188>|": 151857,
|
| 136 |
+
"|<EXTRA_TOKENS_189>|": 151858,
|
| 137 |
+
"|<EXTRA_TOKENS_18>|": 151687,
|
| 138 |
+
"|<EXTRA_TOKENS_190>|": 151859,
|
| 139 |
+
"|<EXTRA_TOKENS_191>|": 151860,
|
| 140 |
+
"|<EXTRA_TOKENS_192>|": 151861,
|
| 141 |
+
"|<EXTRA_TOKENS_193>|": 151862,
|
| 142 |
+
"|<EXTRA_TOKENS_194>|": 151863,
|
| 143 |
+
"|<EXTRA_TOKENS_195>|": 151864,
|
| 144 |
+
"|<EXTRA_TOKENS_196>|": 151865,
|
| 145 |
+
"|<EXTRA_TOKENS_197>|": 151866,
|
| 146 |
+
"|<EXTRA_TOKENS_198>|": 151867,
|
| 147 |
+
"|<EXTRA_TOKENS_199>|": 151868,
|
| 148 |
+
"|<EXTRA_TOKENS_19>|": 151688,
|
| 149 |
+
"|<EXTRA_TOKENS_1>|": 151670,
|
| 150 |
+
"|<EXTRA_TOKENS_200>|": 151869,
|
| 151 |
+
"|<EXTRA_TOKENS_201>|": 151870,
|
| 152 |
+
"|<EXTRA_TOKENS_202>|": 151871,
|
| 153 |
+
"|<EXTRA_TOKENS_203>|": 151872,
|
| 154 |
+
"|<EXTRA_TOKENS_204>|": 151873,
|
| 155 |
+
"|<EXTRA_TOKENS_205>|": 151874,
|
| 156 |
+
"|<EXTRA_TOKENS_206>|": 151875,
|
| 157 |
+
"|<EXTRA_TOKENS_207>|": 151876,
|
| 158 |
+
"|<EXTRA_TOKENS_208>|": 151877,
|
| 159 |
+
"|<EXTRA_TOKENS_209>|": 151878,
|
| 160 |
+
"|<EXTRA_TOKENS_20>|": 151689,
|
| 161 |
+
"|<EXTRA_TOKENS_210>|": 151879,
|
| 162 |
+
"|<EXTRA_TOKENS_211>|": 151880,
|
| 163 |
+
"|<EXTRA_TOKENS_212>|": 151881,
|
| 164 |
+
"|<EXTRA_TOKENS_213>|": 151882,
|
| 165 |
+
"|<EXTRA_TOKENS_214>|": 151883,
|
| 166 |
+
"|<EXTRA_TOKENS_215>|": 151884,
|
| 167 |
+
"|<EXTRA_TOKENS_216>|": 151885,
|
| 168 |
+
"|<EXTRA_TOKENS_217>|": 151886,
|
| 169 |
+
"|<EXTRA_TOKENS_218>|": 151887,
|
| 170 |
+
"|<EXTRA_TOKENS_219>|": 151888,
|
| 171 |
+
"|<EXTRA_TOKENS_21>|": 151690,
|
| 172 |
+
"|<EXTRA_TOKENS_220>|": 151889,
|
| 173 |
+
"|<EXTRA_TOKENS_221>|": 151890,
|
| 174 |
+
"|<EXTRA_TOKENS_222>|": 151891,
|
| 175 |
+
"|<EXTRA_TOKENS_223>|": 151892,
|
| 176 |
+
"|<EXTRA_TOKENS_224>|": 151893,
|
| 177 |
+
"|<EXTRA_TOKENS_225>|": 151894,
|
| 178 |
+
"|<EXTRA_TOKENS_226>|": 151895,
|
| 179 |
+
"|<EXTRA_TOKENS_227>|": 151896,
|
| 180 |
+
"|<EXTRA_TOKENS_228>|": 151897,
|
| 181 |
+
"|<EXTRA_TOKENS_229>|": 151898,
|
| 182 |
+
"|<EXTRA_TOKENS_22>|": 151691,
|
| 183 |
+
"|<EXTRA_TOKENS_230>|": 151899,
|
| 184 |
+
"|<EXTRA_TOKENS_231>|": 151900,
|
| 185 |
+
"|<EXTRA_TOKENS_232>|": 151901,
|
| 186 |
+
"|<EXTRA_TOKENS_233>|": 151902,
|
| 187 |
+
"|<EXTRA_TOKENS_234>|": 151903,
|
| 188 |
+
"|<EXTRA_TOKENS_235>|": 151904,
|
| 189 |
+
"|<EXTRA_TOKENS_236>|": 151905,
|
| 190 |
+
"|<EXTRA_TOKENS_237>|": 151906,
|
| 191 |
+
"|<EXTRA_TOKENS_238>|": 151907,
|
| 192 |
+
"|<EXTRA_TOKENS_239>|": 151908,
|
| 193 |
+
"|<EXTRA_TOKENS_23>|": 151692,
|
| 194 |
+
"|<EXTRA_TOKENS_240>|": 151909,
|
| 195 |
+
"|<EXTRA_TOKENS_241>|": 151910,
|
| 196 |
+
"|<EXTRA_TOKENS_242>|": 151911,
|
| 197 |
+
"|<EXTRA_TOKENS_243>|": 151912,
|
| 198 |
+
"|<EXTRA_TOKENS_244>|": 151913,
|
| 199 |
+
"|<EXTRA_TOKENS_245>|": 151914,
|
| 200 |
+
"|<EXTRA_TOKENS_246>|": 151915,
|
| 201 |
+
"|<EXTRA_TOKENS_247>|": 151916,
|
| 202 |
+
"|<EXTRA_TOKENS_248>|": 151917,
|
| 203 |
+
"|<EXTRA_TOKENS_249>|": 151918,
|
| 204 |
+
"|<EXTRA_TOKENS_24>|": 151693,
|
| 205 |
+
"|<EXTRA_TOKENS_250>|": 151919,
|
| 206 |
+
"|<EXTRA_TOKENS_251>|": 151920,
|
| 207 |
+
"|<EXTRA_TOKENS_252>|": 151921,
|
| 208 |
+
"|<EXTRA_TOKENS_253>|": 151922,
|
| 209 |
+
"|<EXTRA_TOKENS_254>|": 151923,
|
| 210 |
+
"|<EXTRA_TOKENS_255>|": 151924,
|
| 211 |
+
"|<EXTRA_TOKENS_256>|": 151925,
|
| 212 |
+
"|<EXTRA_TOKENS_257>|": 151926,
|
| 213 |
+
"|<EXTRA_TOKENS_258>|": 151927,
|
| 214 |
+
"|<EXTRA_TOKENS_259>|": 151928,
|
| 215 |
+
"|<EXTRA_TOKENS_25>|": 151694,
|
| 216 |
+
"|<EXTRA_TOKENS_260>|": 151929,
|
| 217 |
+
"|<EXTRA_TOKENS_261>|": 151930,
|
| 218 |
+
"|<EXTRA_TOKENS_262>|": 151931,
|
| 219 |
+
"|<EXTRA_TOKENS_263>|": 151932,
|
| 220 |
+
"|<EXTRA_TOKENS_264>|": 151933,
|
| 221 |
+
"|<EXTRA_TOKENS_265>|": 151934,
|
| 222 |
+
"|<EXTRA_TOKENS_266>|": 151935,
|
| 223 |
+
"|<EXTRA_TOKENS_26>|": 151695,
|
| 224 |
+
"|<EXTRA_TOKENS_27>|": 151696,
|
| 225 |
+
"|<EXTRA_TOKENS_28>|": 151697,
|
| 226 |
+
"|<EXTRA_TOKENS_29>|": 151698,
|
| 227 |
+
"|<EXTRA_TOKENS_2>|": 151671,
|
| 228 |
+
"|<EXTRA_TOKENS_30>|": 151699,
|
| 229 |
+
"|<EXTRA_TOKENS_31>|": 151700,
|
| 230 |
+
"|<EXTRA_TOKENS_32>|": 151701,
|
| 231 |
+
"|<EXTRA_TOKENS_33>|": 151702,
|
| 232 |
+
"|<EXTRA_TOKENS_34>|": 151703,
|
| 233 |
+
"|<EXTRA_TOKENS_35>|": 151704,
|
| 234 |
+
"|<EXTRA_TOKENS_36>|": 151705,
|
| 235 |
+
"|<EXTRA_TOKENS_37>|": 151706,
|
| 236 |
+
"|<EXTRA_TOKENS_38>|": 151707,
|
| 237 |
+
"|<EXTRA_TOKENS_39>|": 151708,
|
| 238 |
+
"|<EXTRA_TOKENS_3>|": 151672,
|
| 239 |
+
"|<EXTRA_TOKENS_40>|": 151709,
|
| 240 |
+
"|<EXTRA_TOKENS_41>|": 151710,
|
| 241 |
+
"|<EXTRA_TOKENS_42>|": 151711,
|
| 242 |
+
"|<EXTRA_TOKENS_43>|": 151712,
|
| 243 |
+
"|<EXTRA_TOKENS_44>|": 151713,
|
| 244 |
+
"|<EXTRA_TOKENS_45>|": 151714,
|
| 245 |
+
"|<EXTRA_TOKENS_46>|": 151715,
|
| 246 |
+
"|<EXTRA_TOKENS_47>|": 151716,
|
| 247 |
+
"|<EXTRA_TOKENS_48>|": 151717,
|
| 248 |
+
"|<EXTRA_TOKENS_49>|": 151718,
|
| 249 |
+
"|<EXTRA_TOKENS_4>|": 151673,
|
| 250 |
+
"|<EXTRA_TOKENS_50>|": 151719,
|
| 251 |
+
"|<EXTRA_TOKENS_51>|": 151720,
|
| 252 |
+
"|<EXTRA_TOKENS_52>|": 151721,
|
| 253 |
+
"|<EXTRA_TOKENS_53>|": 151722,
|
| 254 |
+
"|<EXTRA_TOKENS_54>|": 151723,
|
| 255 |
+
"|<EXTRA_TOKENS_55>|": 151724,
|
| 256 |
+
"|<EXTRA_TOKENS_56>|": 151725,
|
| 257 |
+
"|<EXTRA_TOKENS_57>|": 151726,
|
| 258 |
+
"|<EXTRA_TOKENS_58>|": 151727,
|
| 259 |
+
"|<EXTRA_TOKENS_59>|": 151728,
|
| 260 |
+
"|<EXTRA_TOKENS_5>|": 151674,
|
| 261 |
+
"|<EXTRA_TOKENS_60>|": 151729,
|
| 262 |
+
"|<EXTRA_TOKENS_61>|": 151730,
|
| 263 |
+
"|<EXTRA_TOKENS_62>|": 151731,
|
| 264 |
+
"|<EXTRA_TOKENS_63>|": 151732,
|
| 265 |
+
"|<EXTRA_TOKENS_64>|": 151733,
|
| 266 |
+
"|<EXTRA_TOKENS_65>|": 151734,
|
| 267 |
+
"|<EXTRA_TOKENS_66>|": 151735,
|
| 268 |
+
"|<EXTRA_TOKENS_67>|": 151736,
|
| 269 |
+
"|<EXTRA_TOKENS_68>|": 151737,
|
| 270 |
+
"|<EXTRA_TOKENS_69>|": 151738,
|
| 271 |
+
"|<EXTRA_TOKENS_6>|": 151675,
|
| 272 |
+
"|<EXTRA_TOKENS_70>|": 151739,
|
| 273 |
+
"|<EXTRA_TOKENS_71>|": 151740,
|
| 274 |
+
"|<EXTRA_TOKENS_72>|": 151741,
|
| 275 |
+
"|<EXTRA_TOKENS_73>|": 151742,
|
| 276 |
+
"|<EXTRA_TOKENS_74>|": 151743,
|
| 277 |
+
"|<EXTRA_TOKENS_75>|": 151744,
|
| 278 |
+
"|<EXTRA_TOKENS_76>|": 151745,
|
| 279 |
+
"|<EXTRA_TOKENS_77>|": 151746,
|
| 280 |
+
"|<EXTRA_TOKENS_78>|": 151747,
|
| 281 |
+
"|<EXTRA_TOKENS_79>|": 151748,
|
| 282 |
+
"|<EXTRA_TOKENS_7>|": 151676,
|
| 283 |
+
"|<EXTRA_TOKENS_80>|": 151749,
|
| 284 |
+
"|<EXTRA_TOKENS_81>|": 151750,
|
| 285 |
+
"|<EXTRA_TOKENS_82>|": 151751,
|
| 286 |
+
"|<EXTRA_TOKENS_83>|": 151752,
|
| 287 |
+
"|<EXTRA_TOKENS_84>|": 151753,
|
| 288 |
+
"|<EXTRA_TOKENS_85>|": 151754,
|
| 289 |
+
"|<EXTRA_TOKENS_86>|": 151755,
|
| 290 |
+
"|<EXTRA_TOKENS_87>|": 151756,
|
| 291 |
+
"|<EXTRA_TOKENS_88>|": 151757,
|
| 292 |
+
"|<EXTRA_TOKENS_89>|": 151758,
|
| 293 |
+
"|<EXTRA_TOKENS_8>|": 151677,
|
| 294 |
+
"|<EXTRA_TOKENS_90>|": 151759,
|
| 295 |
+
"|<EXTRA_TOKENS_91>|": 151760,
|
| 296 |
+
"|<EXTRA_TOKENS_92>|": 151761,
|
| 297 |
+
"|<EXTRA_TOKENS_93>|": 151762,
|
| 298 |
+
"|<EXTRA_TOKENS_94>|": 151763,
|
| 299 |
+
"|<EXTRA_TOKENS_95>|": 151764,
|
| 300 |
+
"|<EXTRA_TOKENS_96>|": 151765,
|
| 301 |
+
"|<EXTRA_TOKENS_97>|": 151766,
|
| 302 |
+
"|<EXTRA_TOKENS_98>|": 151767,
|
| 303 |
+
"|<EXTRA_TOKENS_99>|": 151768,
|
| 304 |
+
"|<EXTRA_TOKENS_9>|": 151678
|
| 305 |
+
}
|
chat_template.jinja
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{% set DEMO_STYLES = ['point_count','pointing','cosyn_point','user_qa','long_caption','short_caption','video_long_caption','video_short_caption','video_point_track_per_frame','video_point_track_start_end','video_point_track_all_frames','video_single_point_track_start_end','video_transcript','video_clip_caption_start_end','video_clip_caption_start_end_in_seconds','video_clip_transcript_start_end','video_clip_transcript_start_end_in_seconds','video_frame_caption_timestamp','video_frame_caption_timestamp_in_seconds','correction_qa','text_sft','video_point','video_point_count','video_count','video_count_point','multi_image_pointing','multi_image_counting','multi_image_point_then_count','multi_image_count_then_point','demo','a_okvqa_mc','ai2_diagram_no_letter','ai2_diagram','science_qa','multi_image_mc','multi_image_mc_exp','mantis_instruct_mc','video_multiple_choice','video_multiple_choice_count_without_pointing','video_multiple_choice_multiple_correct','video_multiple_choice_w_subtitle'] %}{% set image_count = namespace(value=0) %}{% set video_count = namespace(value=0) %}{% set has_subtitle = messages and messages[0]['role'].lower() == 'subtitle' %}{% for message in messages %}{% if message['content'] is not string %}{% for content in message['content'] %}{% if content['type'] == 'image' or 'image' in content or 'image_url' in content %}{% set image_count.value = image_count.value + 1 %}{% elif content['type'] == 'video' or 'video' in content or 'video_url' in content %}{% set video_count.value = video_count.value + 1 %}{% endif %}{% endfor %}{% endif %}{% endfor %}{% if image_count.value == 1 %}{{ '<|image|>' }}{% elif image_count.value > 1 %}{% for i in range(image_count.value) %}{{ 'Image ' ~ (i + 1) ~ '<|image|>' }}{% endfor %}{% endif %}{% for _ in range(video_count.value) %}{{ '<|video|>' }}{% endfor %}{% if has_subtitle %}{{ messages[0]['content'] }}{% endif %}{% for message in messages %}{% set role = message['role'].lower() %}{% if role == 'subtitle' %}{% continue %}{% endif %}{% set conv_index = loop.index - (1 if has_subtitle else 0) %}{%- if (conv_index % 2 == 1 and role != 'user') or (conv_index % 2 == 0 and role != 'assistant') -%}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{%- endif -%}{% if message['content'] is string %}{% set text_content = message['content'] %}{% else %}{% set m = namespace(text='') %}{% for content in message['content'] %}{% if content['type'] == 'text' %}{% if content['style'] is defined and content['style'] not in DEMO_STYLES %}{% set seg = content['style'] ~ ': ' ~ content['text'] %}{% else %}{% set seg = content['text'] %}{% endif %}{% set m.text = m.text ~ ('' if not m.text else ' ') ~ seg %}{% endif %}{% endfor %}{% set text_content = m.text %}{% endif %}{% if role == 'user' %}User: {{ text_content }}{% else %} {{ text_content }}{% endif %}{% endfor %}{% if add_generation_prompt %} Assistant:{% endif %}
|
config.json
ADDED
|
@@ -0,0 +1,98 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"adapter_config": {
|
| 3 |
+
"attention_dropout": 0.0,
|
| 4 |
+
"attn_implementation": "sdpa",
|
| 5 |
+
"float32_attention": true,
|
| 6 |
+
"head_dim": 72,
|
| 7 |
+
"hidden_act": "silu",
|
| 8 |
+
"hidden_size": 1152,
|
| 9 |
+
"image_feature_dropout": 0.0,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 9728,
|
| 12 |
+
"model_type": "molmo2",
|
| 13 |
+
"num_attention_heads": 16,
|
| 14 |
+
"num_key_value_heads": 16,
|
| 15 |
+
"pooling_attention_mask": true,
|
| 16 |
+
"residual_dropout": 0.0,
|
| 17 |
+
"text_hidden_size": 2560,
|
| 18 |
+
"vit_layers": [
|
| 19 |
+
-3,
|
| 20 |
+
-9
|
| 21 |
+
]
|
| 22 |
+
},
|
| 23 |
+
"architectures": [
|
| 24 |
+
"Molmo2ForConditionalGeneration"
|
| 25 |
+
],
|
| 26 |
+
"auto_map": {
|
| 27 |
+
"AutoConfig": "configuration_molmo2.Molmo2Config",
|
| 28 |
+
"AutoModelForImageTextToText": "modeling_molmo2.Molmo2ForConditionalGeneration"
|
| 29 |
+
},
|
| 30 |
+
"dtype": "float32",
|
| 31 |
+
"frame_end_token_id": 151944,
|
| 32 |
+
"frame_start_token_id": 151943,
|
| 33 |
+
"image_col_id": 151939,
|
| 34 |
+
"image_end_token_id": 151937,
|
| 35 |
+
"image_high_res_id": 151938,
|
| 36 |
+
"image_low_res_id": 151942,
|
| 37 |
+
"image_patch_id": 151938,
|
| 38 |
+
"image_start_token_id": 151936,
|
| 39 |
+
"initializer_range": 0.02,
|
| 40 |
+
"low_res_image_start_token_id": 151940,
|
| 41 |
+
"model_type": "molmo2",
|
| 42 |
+
"text_config": {
|
| 43 |
+
"additional_vocab_size": 128,
|
| 44 |
+
"attention_dropout": 0.0,
|
| 45 |
+
"attn_implementation": "sdpa",
|
| 46 |
+
"embedding_dropout": 0.0,
|
| 47 |
+
"head_dim": 128,
|
| 48 |
+
"hidden_act": "silu",
|
| 49 |
+
"hidden_size": 2560,
|
| 50 |
+
"initializer_range": 0.02,
|
| 51 |
+
"intermediate_size": 9728,
|
| 52 |
+
"layer_norm_eps": 1e-06,
|
| 53 |
+
"max_position_embeddings": 10240,
|
| 54 |
+
"model_type": "molmo2_text",
|
| 55 |
+
"norm_after": false,
|
| 56 |
+
"num_attention_heads": 32,
|
| 57 |
+
"num_hidden_layers": 36,
|
| 58 |
+
"num_key_value_heads": 8,
|
| 59 |
+
"qk_norm_type": "qwen3",
|
| 60 |
+
"qkv_bias": false,
|
| 61 |
+
"residual_dropout": 0.0,
|
| 62 |
+
"rope_scaling": null,
|
| 63 |
+
"rope_scaling_layers": null,
|
| 64 |
+
"rope_theta": 1000000.0,
|
| 65 |
+
"use_cache": true,
|
| 66 |
+
"use_qk_norm": true,
|
| 67 |
+
"vocab_size": 151936
|
| 68 |
+
},
|
| 69 |
+
"tie_word_embeddings": false,
|
| 70 |
+
"transformers_version": "4.47.0",
|
| 71 |
+
"use_cache": true,
|
| 72 |
+
"use_frame_special_tokens": false,
|
| 73 |
+
"vit_config": {
|
| 74 |
+
"attention_dropout": 0.0,
|
| 75 |
+
"attn_implementation": "sdpa",
|
| 76 |
+
"float32_attention": true,
|
| 77 |
+
"head_dim": 72,
|
| 78 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 79 |
+
"hidden_size": 1152,
|
| 80 |
+
"image_default_input_size": [
|
| 81 |
+
378,
|
| 82 |
+
378
|
| 83 |
+
],
|
| 84 |
+
"image_num_pos": 729,
|
| 85 |
+
"image_patch_size": 14,
|
| 86 |
+
"initializer_range": 0.02,
|
| 87 |
+
"intermediate_size": 4304,
|
| 88 |
+
"layer_norm_eps": 1e-06,
|
| 89 |
+
"model_type": "molmo2",
|
| 90 |
+
"num_attention_heads": 16,
|
| 91 |
+
"num_hidden_layers": 27,
|
| 92 |
+
"num_key_value_heads": 16,
|
| 93 |
+
"residual_dropout": 0.0
|
| 94 |
+
},
|
| 95 |
+
"eos_token_id": 151645,
|
| 96 |
+
"bos_token_id": 151645,
|
| 97 |
+
"pad_token_id": 151643
|
| 98 |
+
}
|
config.yaml
ADDED
|
@@ -0,0 +1,542 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
run_name: train_hero_4b_03-14-06-17
|
| 2 |
+
model:
|
| 3 |
+
model_name: molmo
|
| 4 |
+
llm:
|
| 5 |
+
d_model: 2560
|
| 6 |
+
n_heads: 32
|
| 7 |
+
n_kv_heads: 8
|
| 8 |
+
head_dim: 128
|
| 9 |
+
qkv_bias: false
|
| 10 |
+
clip_qkv: null
|
| 11 |
+
n_layers: 36
|
| 12 |
+
mlp_ratio: 4
|
| 13 |
+
mlp_hidden_size: 19456
|
| 14 |
+
activation_type: swiglu
|
| 15 |
+
block_type: sequential
|
| 16 |
+
rope: true
|
| 17 |
+
rope_full_precision: true
|
| 18 |
+
rope_theta: 1000000.0
|
| 19 |
+
rope_type: default
|
| 20 |
+
rope_factor: null
|
| 21 |
+
rope_high_freq_factor: null
|
| 22 |
+
rope_low_freq_factor: null
|
| 23 |
+
rope_original_max_position_embeddings: null
|
| 24 |
+
rope_attention_factor: null
|
| 25 |
+
rope_beta_fast: null
|
| 26 |
+
rope_beta_slow: null
|
| 27 |
+
rope_mscale: null
|
| 28 |
+
rope_mscale_all_dim: null
|
| 29 |
+
rope_truncate: null
|
| 30 |
+
attention_type: sdpa
|
| 31 |
+
full_attention_layers: null
|
| 32 |
+
sliding_attention_rope_scaling: false
|
| 33 |
+
float32_attention: true
|
| 34 |
+
attention_dropout: 0.0
|
| 35 |
+
attention_layer_norm: true
|
| 36 |
+
attention_layer_norm_type: qwen3
|
| 37 |
+
residual_dropout: 0.1
|
| 38 |
+
response_residual_dropout: 0.0
|
| 39 |
+
layer_norm_type: rms
|
| 40 |
+
layer_norm_with_affine: true
|
| 41 |
+
layer_norm_eps: 1.0e-06
|
| 42 |
+
attention_layer_norm_with_affine: true
|
| 43 |
+
max_sequence_length: 10240
|
| 44 |
+
max_position_embeddings: null
|
| 45 |
+
include_bias: false
|
| 46 |
+
bias_for_layer_norm: null
|
| 47 |
+
norm_after: false
|
| 48 |
+
moe_num_experts: 8
|
| 49 |
+
moe_top_k: 2
|
| 50 |
+
moe_mlp_impl: sparse
|
| 51 |
+
moe_log_expert_assignment: false
|
| 52 |
+
moe_shared_expert: false
|
| 53 |
+
moe_lbl_in_fp32: false
|
| 54 |
+
moe_interleave: false
|
| 55 |
+
moe_loss_weight: 0.1
|
| 56 |
+
moe_zloss_weight: null
|
| 57 |
+
moe_dropless: true
|
| 58 |
+
moe_capacity_factor: 1.25
|
| 59 |
+
embedding_dropout: 0.0
|
| 60 |
+
scale_logits: false
|
| 61 |
+
vocab_size: 151936
|
| 62 |
+
additional_vocab_size: 128
|
| 63 |
+
weight_tying: true
|
| 64 |
+
embedding_size: 151936
|
| 65 |
+
use_position_ids: true
|
| 66 |
+
tokenizer:
|
| 67 |
+
identifier: Qwen/Qwen3-4B
|
| 68 |
+
tokenizer_dir: null
|
| 69 |
+
init_path: /weka/oe-training-default/mm-olmo/pretrained_llms/qwen3-4b.pt
|
| 70 |
+
init_incremental: null
|
| 71 |
+
new_embedding_init_range: 0.02
|
| 72 |
+
initializer_range: 0.02
|
| 73 |
+
normalize_input_embeds: false
|
| 74 |
+
activation_checkpoint: whole_layer
|
| 75 |
+
compile: blocks
|
| 76 |
+
fix_pad_tokenizer: false
|
| 77 |
+
init_std: 0.02
|
| 78 |
+
init_fn: normal
|
| 79 |
+
init_cutoff_factor: null
|
| 80 |
+
vision_backbone:
|
| 81 |
+
vit:
|
| 82 |
+
image_model_type: siglip
|
| 83 |
+
image_default_input_size:
|
| 84 |
+
- 378
|
| 85 |
+
- 378
|
| 86 |
+
image_patch_size: 14
|
| 87 |
+
image_pos_patch_size: 14
|
| 88 |
+
image_emb_dim: 1152
|
| 89 |
+
image_num_heads: 16
|
| 90 |
+
image_num_key_value_heads: 16
|
| 91 |
+
image_num_layers: 27
|
| 92 |
+
image_head_dim: 72
|
| 93 |
+
image_mlp_dim: 4304
|
| 94 |
+
image_mlp_activations: gelu_pytorch_tanh
|
| 95 |
+
image_dropout_rate: 0.0
|
| 96 |
+
image_num_pos: 729
|
| 97 |
+
image_norm_eps: 1.0e-06
|
| 98 |
+
attention_dropout: 0.0
|
| 99 |
+
residual_dropout: 0.0
|
| 100 |
+
initializer_range: 0.02
|
| 101 |
+
float32_attention: true
|
| 102 |
+
attention_type: sdpa
|
| 103 |
+
sdpa_backend: all
|
| 104 |
+
activation_checkpointing: true
|
| 105 |
+
init_path: /weka/oe-training-default/mm-olmo/pretrained_image_encoders/siglip2-so400m-14-384.pt
|
| 106 |
+
resize_mode: siglip
|
| 107 |
+
pad_value: 0.0
|
| 108 |
+
normalize: siglip
|
| 109 |
+
image_pooling_2d: attention_meanq
|
| 110 |
+
pooling_attention_mask: true
|
| 111 |
+
image_projector: mlp
|
| 112 |
+
image_padding_embed: null
|
| 113 |
+
vit_layers:
|
| 114 |
+
- -3
|
| 115 |
+
- -9
|
| 116 |
+
skip_unused_layers: true
|
| 117 |
+
use_deepstack: false
|
| 118 |
+
share_connector: false
|
| 119 |
+
image_feature_dropout: 0.0
|
| 120 |
+
connector_activation_checkpointing: true
|
| 121 |
+
compile_vit: blocks
|
| 122 |
+
pool_size_embeds: null
|
| 123 |
+
compile_connector: dynamic
|
| 124 |
+
normalize_on_gpu: true
|
| 125 |
+
data_formatter:
|
| 126 |
+
prompt_templates: uber_model
|
| 127 |
+
message_format: role
|
| 128 |
+
system_prompt: demo_or_style
|
| 129 |
+
always_start_with_space: false
|
| 130 |
+
default_inference_len: 65
|
| 131 |
+
select_answer: best
|
| 132 |
+
debug: false
|
| 133 |
+
image_last: false
|
| 134 |
+
format_message_list: null
|
| 135 |
+
p_one_message: 0.0
|
| 136 |
+
eval_system_prompt_mapping: null
|
| 137 |
+
p_choice_content_in_mc: 1.0
|
| 138 |
+
template_video_mc_questions: true
|
| 139 |
+
pointing_format: html-v2
|
| 140 |
+
points_decimal_places: 1
|
| 141 |
+
use_seperate_non_pointing_qa_style: false
|
| 142 |
+
timestamp_mode: 50-percent-seconds
|
| 143 |
+
output_timestamp_mode: seconds
|
| 144 |
+
seconds_decimal_places: 1
|
| 145 |
+
p_multi_point_all_image: 0.0
|
| 146 |
+
use_seperate_count_without_pointing_style: false
|
| 147 |
+
sample_random_initial_point: true
|
| 148 |
+
mm_preprocessor:
|
| 149 |
+
crop_mode: overlap-and-resize-c2
|
| 150 |
+
use_col_tokens: true
|
| 151 |
+
max_crops: 8
|
| 152 |
+
high_res_max_crops: 24
|
| 153 |
+
p_high_res: 0.0
|
| 154 |
+
pooling_w: 2
|
| 155 |
+
pooling_h: 2
|
| 156 |
+
overlap_margins:
|
| 157 |
+
- 4
|
| 158 |
+
- 4
|
| 159 |
+
max_images: null
|
| 160 |
+
max_multi_image_crops: 4
|
| 161 |
+
multi_image_pooling_w: 2
|
| 162 |
+
multi_image_pooling_h: 2
|
| 163 |
+
use_single_crop_col_tokens: null
|
| 164 |
+
use_single_crop_start_token: false
|
| 165 |
+
max_answer_len: null
|
| 166 |
+
last_message_loss_only: false
|
| 167 |
+
max_text_tokens: null
|
| 168 |
+
loss_token_weighting: root_subsegments
|
| 169 |
+
image_padding_mask: false
|
| 170 |
+
legacy_image_mask: false
|
| 171 |
+
bi_directional_attn: null
|
| 172 |
+
parallelism:
|
| 173 |
+
data_parallel_replicate_degree: 1
|
| 174 |
+
enable_compiled_autograd: false
|
| 175 |
+
data_parallel_shard_degree: -1
|
| 176 |
+
fsdp_reshard_after_forward: default
|
| 177 |
+
context_parallel_config:
|
| 178 |
+
degree: 1
|
| 179 |
+
attention_type: ulysses
|
| 180 |
+
load_balancer: ulysses
|
| 181 |
+
head_stride: 1
|
| 182 |
+
tensor_parallel_config:
|
| 183 |
+
degree: 1
|
| 184 |
+
enable_async: false
|
| 185 |
+
data_parallel_config:
|
| 186 |
+
name: fsdp
|
| 187 |
+
param_dtype: null
|
| 188 |
+
reduce_dtype: float32
|
| 189 |
+
num_replicas: null
|
| 190 |
+
shard_degree: null
|
| 191 |
+
wrapping_strategy: full
|
| 192 |
+
prefetch_factor: 0
|
| 193 |
+
context_parallel_rotate_method: allgather
|
| 194 |
+
seed: 6198
|
| 195 |
+
epoch: null
|
| 196 |
+
dry_run: false
|
| 197 |
+
ft_llm: true
|
| 198 |
+
ft_vit: true
|
| 199 |
+
ft_connector: true
|
| 200 |
+
ft_embedding: lm_head
|
| 201 |
+
optimizer:
|
| 202 |
+
name: adamw
|
| 203 |
+
learning_rate: 0.0001
|
| 204 |
+
weight_decay: 0.01
|
| 205 |
+
betas:
|
| 206 |
+
- 0.9
|
| 207 |
+
- 0.95
|
| 208 |
+
eps: 1.0e-05
|
| 209 |
+
connector_learning_rate: 5.0e-06
|
| 210 |
+
vit_learning_rate: 5.0e-06
|
| 211 |
+
llm_learning_rate: 1.0e-05
|
| 212 |
+
frame_selector_learning_rate: 0.0001
|
| 213 |
+
temporal_token_scorer_learning_rate: 0.0001
|
| 214 |
+
connector_weight_decay: 0.0
|
| 215 |
+
vit_weight_decay: 0.0
|
| 216 |
+
llm_weight_decay: 0.0
|
| 217 |
+
frame_selector_weight_decay: 0.01
|
| 218 |
+
temporal_token_scorer_weight_decay: 0.01
|
| 219 |
+
connector_betas:
|
| 220 |
+
- 0.9
|
| 221 |
+
- 0.95
|
| 222 |
+
vit_betas:
|
| 223 |
+
- 0.9
|
| 224 |
+
- 0.95
|
| 225 |
+
llm_betas:
|
| 226 |
+
- 0.9
|
| 227 |
+
- 0.95
|
| 228 |
+
frame_selector_betas:
|
| 229 |
+
- 0.9
|
| 230 |
+
- 0.95
|
| 231 |
+
temporal_token_scorer_betas:
|
| 232 |
+
- 0.9
|
| 233 |
+
- 0.95
|
| 234 |
+
connector_eps: 1.0e-06
|
| 235 |
+
vit_eps: 1.0e-06
|
| 236 |
+
llm_eps: 1.0e-06
|
| 237 |
+
frame_selector_eps: 1.0e-06
|
| 238 |
+
temporal_token_scorer_eps: 1.0e-06
|
| 239 |
+
metrics_log_interval: -1
|
| 240 |
+
scheduler:
|
| 241 |
+
name: multimodal
|
| 242 |
+
units: steps
|
| 243 |
+
t_warmup: 100
|
| 244 |
+
t_max: null
|
| 245 |
+
alpha_f: 0.1
|
| 246 |
+
connector_t_warmup: 200
|
| 247 |
+
vit_t_warmup: 200
|
| 248 |
+
llm_t_warmup: 200
|
| 249 |
+
frame_selector_t_warmup: 200
|
| 250 |
+
temporal_token_scorer_t_warmup: 200
|
| 251 |
+
grad_clip_warmup_steps: null
|
| 252 |
+
grad_clip_warmup_factor: null
|
| 253 |
+
warmup_min_lr: 0.0
|
| 254 |
+
data:
|
| 255 |
+
dataset: null
|
| 256 |
+
mixture: null
|
| 257 |
+
root_size_mixture: null
|
| 258 |
+
kwargs_mixture:
|
| 259 |
+
- rate: 0.2
|
| 260 |
+
datasets:
|
| 261 |
+
- dataset_name: webolmoSyntheticGround__v0__template
|
| 262 |
+
sampling_rate: null
|
| 263 |
+
root_size_factor: null
|
| 264 |
+
message_weight: null
|
| 265 |
+
override_p_high_res: null
|
| 266 |
+
- dataset_name: webolmoSyntheticGround__v0__gpt
|
| 267 |
+
sampling_rate: null
|
| 268 |
+
root_size_factor: null
|
| 269 |
+
message_weight: null
|
| 270 |
+
override_p_high_res: null
|
| 271 |
+
- dataset_name: pixmo_points_single_web
|
| 272 |
+
sampling_rate: null
|
| 273 |
+
root_size_factor: null
|
| 274 |
+
message_weight: null
|
| 275 |
+
override_p_high_res: null
|
| 276 |
+
- dataset_name: screenshot_qa
|
| 277 |
+
sampling_rate: null
|
| 278 |
+
root_size_factor: null
|
| 279 |
+
message_weight: null
|
| 280 |
+
override_p_high_res: null
|
| 281 |
+
name: null
|
| 282 |
+
- rate: 0.05
|
| 283 |
+
datasets:
|
| 284 |
+
- dataset_name: webolmoSynthetic__train_gemini_3_v0_like_combined_postprocessed_version2__weighted_221__random_gaussian__molmo_web_think__steps_10
|
| 285 |
+
sampling_rate: null
|
| 286 |
+
root_size_factor: null
|
| 287 |
+
message_weight: null
|
| 288 |
+
override_p_high_res: null
|
| 289 |
+
name: null
|
| 290 |
+
- rate: 0.1
|
| 291 |
+
datasets:
|
| 292 |
+
- dataset_name: webolmoSynthetic__gemini_webvoyager_like_19k_feb20_version2__weighted_221__random_gaussian__molmo_web_think__steps_10
|
| 293 |
+
sampling_rate: null
|
| 294 |
+
root_size_factor: null
|
| 295 |
+
message_weight: null
|
| 296 |
+
override_p_high_res: null
|
| 297 |
+
name: null
|
| 298 |
+
- rate: 0.2
|
| 299 |
+
datasets:
|
| 300 |
+
- dataset_name: webolmoSynthetic__gemini_om2w_combined_33k__weighted_221__random_gaussian__molmo_web_think__steps_10
|
| 301 |
+
sampling_rate: null
|
| 302 |
+
root_size_factor: null
|
| 303 |
+
message_weight: null
|
| 304 |
+
override_p_high_res: null
|
| 305 |
+
name: null
|
| 306 |
+
- rate: 0.18
|
| 307 |
+
datasets:
|
| 308 |
+
- dataset_name: webolmoSynthetic__heuristic_filtered_multi_agent_combined_version2__goal__random_gaussian__molmo_web_think__steps_10
|
| 309 |
+
sampling_rate: null
|
| 310 |
+
root_size_factor: null
|
| 311 |
+
message_weight: null
|
| 312 |
+
override_p_high_res: null
|
| 313 |
+
name: null
|
| 314 |
+
- rate: 0.18
|
| 315 |
+
datasets:
|
| 316 |
+
- dataset_name: snorkel_0312_with_gemini_thoughts__weighted_1111__random_gaussian__molmo_web_think__steps_10
|
| 317 |
+
sampling_rate: null
|
| 318 |
+
root_size_factor: null
|
| 319 |
+
message_weight: null
|
| 320 |
+
override_p_high_res: null
|
| 321 |
+
name: null
|
| 322 |
+
- rate: 0.02
|
| 323 |
+
datasets:
|
| 324 |
+
- dataset_name: webolmoSynthetic__atomic_actions_find_and_open_successful__goal__random_gaussian__molmo_web_think__steps_10
|
| 325 |
+
sampling_rate: null
|
| 326 |
+
root_size_factor: null
|
| 327 |
+
message_weight: null
|
| 328 |
+
override_p_high_res: null
|
| 329 |
+
- dataset_name: webolmoSynthetic__atomic_actions_fill_form_successful__goal__random_gaussian__molmo_web_think__steps_10
|
| 330 |
+
sampling_rate: null
|
| 331 |
+
root_size_factor: null
|
| 332 |
+
message_weight: null
|
| 333 |
+
override_p_high_res: null
|
| 334 |
+
name: null
|
| 335 |
+
- rate: 0.05
|
| 336 |
+
datasets:
|
| 337 |
+
- dataset_name: snorkel_0312_STEPS_with_gemini_thoughts__goal__random_gaussian__molmo_web_think__steps_10
|
| 338 |
+
sampling_rate: null
|
| 339 |
+
root_size_factor: null
|
| 340 |
+
message_weight: null
|
| 341 |
+
override_p_high_res: null
|
| 342 |
+
name: null
|
| 343 |
+
- rate: 0.02
|
| 344 |
+
datasets:
|
| 345 |
+
- dataset_name: webolmoSynthetic__node_traversal_successful_ML_scroll_100_720_1280__weighted_221__random_gaussian__molmo_web_think__steps_10
|
| 346 |
+
sampling_rate: null
|
| 347 |
+
root_size_factor: null
|
| 348 |
+
message_weight: null
|
| 349 |
+
override_p_high_res: null
|
| 350 |
+
name: null
|
| 351 |
+
split: train
|
| 352 |
+
seed: 50189
|
| 353 |
+
pad: to_max
|
| 354 |
+
sequence_length: 10240
|
| 355 |
+
max_text_seq_len: null
|
| 356 |
+
shuffle: true
|
| 357 |
+
start_index: 0
|
| 358 |
+
packing: null
|
| 359 |
+
enable_variable_sized_token_pooling: true
|
| 360 |
+
num_workers: 2
|
| 361 |
+
drop_last: true
|
| 362 |
+
pin_memory: true
|
| 363 |
+
prefetch_factor: 4
|
| 364 |
+
persistent_workers: false
|
| 365 |
+
timeout: 0
|
| 366 |
+
restore_dataloader: true
|
| 367 |
+
fast_forward_batches: null
|
| 368 |
+
evaluators: []
|
| 369 |
+
eval_interval: 50000
|
| 370 |
+
inf_evaluators:
|
| 371 |
+
- label: webolmoSynthetic__train_gemini_3_v0_like_combined_postprocessed__HL__random_gaussian__molmo_web_think__steps_10
|
| 372 |
+
data:
|
| 373 |
+
dataset: webolmoSynthetic__train_gemini_3_v0_like_combined_postprocessed__HL__random_gaussian__molmo_web_think__steps_10
|
| 374 |
+
mixture: null
|
| 375 |
+
root_size_mixture: null
|
| 376 |
+
kwargs_mixture: null
|
| 377 |
+
split: val
|
| 378 |
+
seed: 691203
|
| 379 |
+
pad: to_max
|
| 380 |
+
sequence_length: 10240
|
| 381 |
+
max_text_seq_len: null
|
| 382 |
+
shuffle: true
|
| 383 |
+
start_index: 0
|
| 384 |
+
packing: null
|
| 385 |
+
enable_variable_sized_token_pooling: true
|
| 386 |
+
num_workers: 2
|
| 387 |
+
drop_last: true
|
| 388 |
+
pin_memory: true
|
| 389 |
+
prefetch_factor: 4
|
| 390 |
+
persistent_workers: true
|
| 391 |
+
timeout: 0
|
| 392 |
+
evaluator:
|
| 393 |
+
n_to_log: 0
|
| 394 |
+
num_wandb_examples: 32
|
| 395 |
+
save_predictions: null
|
| 396 |
+
save_tokens: false
|
| 397 |
+
vqa_eval: ''
|
| 398 |
+
pointing_eval: false
|
| 399 |
+
point_bench_eval: false
|
| 400 |
+
count_eval: false
|
| 401 |
+
point_count_eval: false
|
| 402 |
+
android_eval: false
|
| 403 |
+
clock_eval: false
|
| 404 |
+
clock_bench_eval: false
|
| 405 |
+
math_vista_eval: false
|
| 406 |
+
temp_compass_eval: ''
|
| 407 |
+
temp_compass_disable_api: false
|
| 408 |
+
video_mme_eval: ''
|
| 409 |
+
mme_videoocr_eval: false
|
| 410 |
+
mlvu_gen_eval: false
|
| 411 |
+
lvbench_eval: false
|
| 412 |
+
long_video_bench_eval: false
|
| 413 |
+
plm_fgqa_eval: false
|
| 414 |
+
video_hallucer: false
|
| 415 |
+
long_video_bench_caption_eval: false
|
| 416 |
+
vinoground_eval: false
|
| 417 |
+
vixmo_caption_eval: false
|
| 418 |
+
vixmo_caption_eval2: false
|
| 419 |
+
dream1k_caption_eval: false
|
| 420 |
+
vixmo_point_count_eval: false
|
| 421 |
+
vixmo_point_eval: false
|
| 422 |
+
video_object_tracking_eval: ''
|
| 423 |
+
video_single_point_prediction: ''
|
| 424 |
+
video_point_tracking_eval: ''
|
| 425 |
+
refexp_eval: false
|
| 426 |
+
coco_caption_eval: false
|
| 427 |
+
qv_highlights_eval: false
|
| 428 |
+
tomato: false
|
| 429 |
+
temporal_bench: false
|
| 430 |
+
open_qa_eval: false
|
| 431 |
+
mmiu_eval: false
|
| 432 |
+
mulset_eval: false
|
| 433 |
+
ego3d_bench_eval: false
|
| 434 |
+
vsi_bench_eval: false
|
| 435 |
+
uground_eval: false
|
| 436 |
+
web_ground_eval: false
|
| 437 |
+
web_trajs_eval: true
|
| 438 |
+
screenshot_qa_eval: false
|
| 439 |
+
websrc_eval: false
|
| 440 |
+
max_new_tokens: 1024
|
| 441 |
+
device_batch_size: 2
|
| 442 |
+
sampling:
|
| 443 |
+
temperature: 0.0
|
| 444 |
+
top_p: 1.0
|
| 445 |
+
top_k: null
|
| 446 |
+
ngram_size: null
|
| 447 |
+
repetition_penalty: null
|
| 448 |
+
frequency_penalty: null
|
| 449 |
+
subset_num_batches: null
|
| 450 |
+
max_examples: 512
|
| 451 |
+
console_log_interval: 20
|
| 452 |
+
include_image: false
|
| 453 |
+
inf_eval_interval: 50000
|
| 454 |
+
eval_on_last_step: true
|
| 455 |
+
eval_on_load: false
|
| 456 |
+
eval_on: []
|
| 457 |
+
save_folder: /weka/oe-training-default/webolmo/zixianm/checkpoints/train_hero_4b_03-14-06-17
|
| 458 |
+
checkpointer_config:
|
| 459 |
+
save_thread_count: null
|
| 460 |
+
load_thread_count: null
|
| 461 |
+
pre_download: false
|
| 462 |
+
work_dir: null
|
| 463 |
+
throttle_uploads: false
|
| 464 |
+
canceled_check_interval: 50
|
| 465 |
+
save_interval: 1000
|
| 466 |
+
save_at: null
|
| 467 |
+
save_final_optim: false
|
| 468 |
+
save_num_checkpoints_to_keep: 31
|
| 469 |
+
save_final_unsharded_checkpoint: false
|
| 470 |
+
save_interval_ephemeral: null
|
| 471 |
+
save_overwrite: true
|
| 472 |
+
load_path: null
|
| 473 |
+
reset_optimizer_state: false
|
| 474 |
+
reset_trainer_state: false
|
| 475 |
+
initial_model_checkpoint: /weka/oe-training-default/sanghol/molmo/models/uber-v1/uber3.4-synthetic-siglip2-qwen3_4b/step30000
|
| 476 |
+
allow_resume: true
|
| 477 |
+
max_duration: 50000
|
| 478 |
+
global_train_batch_size: 128
|
| 479 |
+
device_train_microbatch_size: 4
|
| 480 |
+
max_grad_norm: 1.0
|
| 481 |
+
multi_component_grad_norm: true
|
| 482 |
+
batch_divisor: global_batch
|
| 483 |
+
max_grad_norm_ratio: null
|
| 484 |
+
precision: amp_bf16
|
| 485 |
+
wandb:
|
| 486 |
+
project: zixianm_webolmo
|
| 487 |
+
entity: prior-ai2
|
| 488 |
+
group: null
|
| 489 |
+
name: train_hero_4b_03-14-06-17
|
| 490 |
+
tags:
|
| 491 |
+
- watching
|
| 492 |
+
log_artifacts: false
|
| 493 |
+
rank_zero_only: true
|
| 494 |
+
log_interval: 20
|
| 495 |
+
allow_resume: true
|
| 496 |
+
finish_on_sigterm: true
|
| 497 |
+
beaker_log_interval: 50
|
| 498 |
+
speed_monitor:
|
| 499 |
+
window_size: 20
|
| 500 |
+
gpu_flops_available: null
|
| 501 |
+
console_log_interval: 20
|
| 502 |
+
enable_timing_logs: false
|
| 503 |
+
gen1_gc_interval: 1
|
| 504 |
+
compile: null
|
| 505 |
+
activation_checkpointing: true
|
| 506 |
+
fsdp:
|
| 507 |
+
fsdp2: true
|
| 508 |
+
precision: float
|
| 509 |
+
use_orig_params: true
|
| 510 |
+
wrapping_strategy: by_block_and_size
|
| 511 |
+
sharding_strategy: FULL_SHARD
|
| 512 |
+
hybrid_sharding_num_model_replicas: null
|
| 513 |
+
softmax_auxiliary_loss: true
|
| 514 |
+
softmax_auxiliary_loss_scale: 0.0001
|
| 515 |
+
response_logits_only: true
|
| 516 |
+
saliency_score_loss_wt: null
|
| 517 |
+
frame_score_loss_wt: null
|
| 518 |
+
frame_score_loss_type: mse
|
| 519 |
+
frame_score_loss_target: 0.7
|
| 520 |
+
time_limit: null
|
| 521 |
+
extra_steps_after_cancel: 0
|
| 522 |
+
python_profiling: false
|
| 523 |
+
torch_profiling: false
|
| 524 |
+
stop_at: 50000
|
| 525 |
+
stop_after: null
|
| 526 |
+
fused_loss: false
|
| 527 |
+
compile_loss: true
|
| 528 |
+
runtime_data:
|
| 529 |
+
args: /gantry-runtime/launch_scripts/train_multitask_model.py hero /weka/oe-training-default/sanghol/molmo/models/uber-v1/uber3.4-synthetic-siglip2-qwen3_4b/step30000
|
| 530 |
+
--save_folder=/weka/oe-training-default/webolmo/zixianm/checkpoints/train_hero_4b_03-14-06-17
|
| 531 |
+
--run_name=train_hero_4b_03-14-06-17 --global_batch_size 128 --device_train_batch_size
|
| 532 |
+
2 --device_eval_batch_size 2 --device_inf_batch_size 2 --duration 50000 --action_token_weight
|
| 533 |
+
1 --num_checkpoints_to_keep 31 --max_crops 8 --seq_len 10240 --save_interval 1000
|
| 534 |
+
--eval_interval 50000 --inf_eval_interval 50000 --save_overwrite
|
| 535 |
+
hostname: jupiter-cs-aus-115.reviz.ai2.in
|
| 536 |
+
date: 03/16/2026, 22:07
|
| 537 |
+
world_size: 64
|
| 538 |
+
resuming_from: /weka/oe-training-default/webolmo/zixianm/checkpoints/train_hero_4b_03-14-06-17/step32000
|
| 539 |
+
beaker_experiment_id: 01KKNFX4X43ASSGM27F63TFDN2
|
| 540 |
+
beaker_experiment_url: null
|
| 541 |
+
wandb_id: jym0efjn
|
| 542 |
+
wandb_url: https://wandb.ai/prior-ai2/zixianm_webolmo/runs/jym0efjn
|
configuration_molmo2.py
ADDED
|
@@ -0,0 +1,390 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Molmo2 configuration
|
| 3 |
+
"""
|
| 4 |
+
|
| 5 |
+
from typing import Optional, Any
|
| 6 |
+
|
| 7 |
+
from transformers import PretrainedConfig
|
| 8 |
+
from transformers.modeling_rope_utils import rope_config_validation
|
| 9 |
+
from transformers.utils import logging
|
| 10 |
+
|
| 11 |
+
logger = logging.get_logger(__name__)
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
class Molmo2VitConfig(PretrainedConfig):
|
| 15 |
+
r"""
|
| 16 |
+
This is the configuration class to store the configuration of a [`Molmo2VisionTransformer`].
|
| 17 |
+
It is used to instantiate a `Molmo2VisionTransformer` according to the specified arguments,
|
| 18 |
+
defining the model architecture.
|
| 19 |
+
|
| 20 |
+
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
|
| 21 |
+
documentation from [`PretrainedConfig`] for more information.
|
| 22 |
+
|
| 23 |
+
Example:
|
| 24 |
+
```python
|
| 25 |
+
>>> from transformers import Molmo2VitConfig, Molmo2VisionTransformer
|
| 26 |
+
|
| 27 |
+
>>> # Initializing a Molmo2VitConfig
|
| 28 |
+
>>> configuration = Molmo2VitConfig()
|
| 29 |
+
|
| 30 |
+
>>> # Initializing a Molmo2VisionTransformer (with random weights)
|
| 31 |
+
>>> model = Molmo2VisionTransformer(configuration)
|
| 32 |
+
|
| 33 |
+
>>> # Accessing the model configuration
|
| 34 |
+
>>> configuration = model.config
|
| 35 |
+
```"""
|
| 36 |
+
|
| 37 |
+
model_type = "molmo2"
|
| 38 |
+
base_config_key = "vit_config"
|
| 39 |
+
|
| 40 |
+
def __init__(
|
| 41 |
+
self,
|
| 42 |
+
hidden_size: int = 1152,
|
| 43 |
+
intermediate_size: int = 4304,
|
| 44 |
+
num_hidden_layers: int = 27,
|
| 45 |
+
num_attention_heads: int = 16,
|
| 46 |
+
num_key_value_heads: int = 16,
|
| 47 |
+
head_dim: int = 72,
|
| 48 |
+
hidden_act: str = "gelu_pytorch_tanh",
|
| 49 |
+
layer_norm_eps: float = 1e-6,
|
| 50 |
+
image_default_input_size: tuple[int, int] = (378, 378),
|
| 51 |
+
image_patch_size: int = 14,
|
| 52 |
+
image_num_pos: int = 577,
|
| 53 |
+
attention_dropout: float = 0.0,
|
| 54 |
+
residual_dropout: float = 0.0,
|
| 55 |
+
initializer_range: float = 0.02,
|
| 56 |
+
float32_attention: bool = True,
|
| 57 |
+
attn_implementation: str = "eager",
|
| 58 |
+
**kwargs,
|
| 59 |
+
):
|
| 60 |
+
self.attn_implementation = attn_implementation
|
| 61 |
+
super().__init__(
|
| 62 |
+
attn_implementation=attn_implementation,
|
| 63 |
+
**kwargs
|
| 64 |
+
)
|
| 65 |
+
self.hidden_size = hidden_size
|
| 66 |
+
self.intermediate_size = intermediate_size
|
| 67 |
+
self.num_hidden_layers = num_hidden_layers
|
| 68 |
+
self.num_attention_heads = num_attention_heads
|
| 69 |
+
self.num_key_value_heads = num_key_value_heads
|
| 70 |
+
self.head_dim = head_dim
|
| 71 |
+
self.hidden_act = hidden_act
|
| 72 |
+
self.layer_norm_eps = layer_norm_eps
|
| 73 |
+
self.image_default_input_size = image_default_input_size
|
| 74 |
+
self.image_patch_size = image_patch_size
|
| 75 |
+
self.image_num_pos = image_num_pos
|
| 76 |
+
self.attention_dropout = attention_dropout
|
| 77 |
+
self.residual_dropout = residual_dropout
|
| 78 |
+
self.initializer_range = initializer_range
|
| 79 |
+
self.float32_attention = float32_attention
|
| 80 |
+
|
| 81 |
+
@property
|
| 82 |
+
def image_num_patch(self):
|
| 83 |
+
h, w = self.image_default_input_size
|
| 84 |
+
return h // self.image_patch_size, w // self.image_patch_size
|
| 85 |
+
|
| 86 |
+
class Molmo2AdapterConfig(PretrainedConfig):
|
| 87 |
+
r"""
|
| 88 |
+
This is the configuration class to store the configuration of Molmo2Adapter. With Molmo2VitConfig,
|
| 89 |
+
It is used to instantiate an Molmo2VisionBackbone according to the specified arguments,
|
| 90 |
+
defining the model architecture.
|
| 91 |
+
|
| 92 |
+
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
|
| 93 |
+
documentation from [`PretrainedConfig`] for more information.
|
| 94 |
+
|
| 95 |
+
Example:
|
| 96 |
+
|
| 97 |
+
```python
|
| 98 |
+
>>> from transformers import Molmo2VitConfig, Molmo2AdapterConfig, Molmo2VisionBackbone
|
| 99 |
+
|
| 100 |
+
>>> # Initializing a Molmo2VitConfig and a Molmo2AdapterConfig
|
| 101 |
+
>>> vit_config = Molmo2VitConfig()
|
| 102 |
+
>>> adapter_config = MolmoPoolingConfig()
|
| 103 |
+
|
| 104 |
+
>>> # Initializing a Molmo2VisionBackbone (with random weights)
|
| 105 |
+
>>> model = Molmo2VisionBackbone(vit_config, adapter_config)
|
| 106 |
+
|
| 107 |
+
>>> # Accessing the model configuration
|
| 108 |
+
>>> vit_configuration = model.vit_config
|
| 109 |
+
>>> adapter_configuration = model.adapter_config
|
| 110 |
+
```"""
|
| 111 |
+
|
| 112 |
+
model_type = "molmo2"
|
| 113 |
+
base_config_key = "adapter_config"
|
| 114 |
+
|
| 115 |
+
def __init__(
|
| 116 |
+
self,
|
| 117 |
+
vit_layers: tuple = (-3, -9),
|
| 118 |
+
pooling_attention_mask: bool = False,
|
| 119 |
+
hidden_size: int = 1152,
|
| 120 |
+
num_attention_heads: int = 16,
|
| 121 |
+
num_key_value_heads: int = 16,
|
| 122 |
+
head_dim: int = 72,
|
| 123 |
+
float32_attention: bool = True,
|
| 124 |
+
attention_dropout: float = 0.0,
|
| 125 |
+
residual_dropout: float = 0.0,
|
| 126 |
+
hidden_act: str = "silu",
|
| 127 |
+
intermediate_size: int = 18944,
|
| 128 |
+
text_hidden_size: int = 3584,
|
| 129 |
+
image_feature_dropout: float = 0.0,
|
| 130 |
+
initializer_range: float = 0.02,
|
| 131 |
+
attn_implementation: str = "eager",
|
| 132 |
+
**kwargs,
|
| 133 |
+
):
|
| 134 |
+
self.attn_implementation = attn_implementation
|
| 135 |
+
super().__init__(
|
| 136 |
+
attn_implementation=attn_implementation,
|
| 137 |
+
**kwargs
|
| 138 |
+
)
|
| 139 |
+
self.vit_layers = vit_layers
|
| 140 |
+
self.pooling_attention_mask = pooling_attention_mask
|
| 141 |
+
self.hidden_size = hidden_size
|
| 142 |
+
self.num_attention_heads = num_attention_heads
|
| 143 |
+
self.num_key_value_heads = num_key_value_heads
|
| 144 |
+
self.head_dim = head_dim
|
| 145 |
+
self.float32_attention = float32_attention
|
| 146 |
+
self.attention_dropout = attention_dropout
|
| 147 |
+
self.residual_dropout = residual_dropout
|
| 148 |
+
self.hidden_act = hidden_act
|
| 149 |
+
self.intermediate_size = intermediate_size
|
| 150 |
+
self.text_hidden_size = text_hidden_size
|
| 151 |
+
self.image_feature_dropout = image_feature_dropout
|
| 152 |
+
self.initializer_range = initializer_range
|
| 153 |
+
|
| 154 |
+
|
| 155 |
+
class Molmo2TextConfig(PretrainedConfig):
|
| 156 |
+
r"""
|
| 157 |
+
This is the configuration class to store the configuration of a [`Molmo2TextModel`]. It is used to instantiate a
|
| 158 |
+
`Molmo2TextModel` according to the specified arguments, defining the model architecture.
|
| 159 |
+
|
| 160 |
+
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
|
| 161 |
+
documentation from [`PretrainedConfig`] for more information.
|
| 162 |
+
|
| 163 |
+
Example:
|
| 164 |
+
```python
|
| 165 |
+
>>> from transformers import Molmo2TextConfig, Molmo2TextModel
|
| 166 |
+
|
| 167 |
+
>>> # Initializing a Molmo2TextConfig
|
| 168 |
+
>>> configuration = Molmo2TextConfig()
|
| 169 |
+
|
| 170 |
+
>>> # Initializing a Molmo2TextModel (with random weights)
|
| 171 |
+
>>> model = Molmo2TextModel(configuration)
|
| 172 |
+
|
| 173 |
+
>>> # Accessing the model configuration
|
| 174 |
+
>>> configuration = model.config
|
| 175 |
+
```"""
|
| 176 |
+
|
| 177 |
+
model_type = "molmo2_text"
|
| 178 |
+
base_config_key = "text_config"
|
| 179 |
+
keys_to_ignore_at_inference = ["past_key_values"]
|
| 180 |
+
base_model_tp_plan = {
|
| 181 |
+
"blocks.*.self_attn.att_proj": "colwise",
|
| 182 |
+
"blocks.*.self_attn.attn_out": "rowwise",
|
| 183 |
+
"blocks.*.mlp.ff_proj": "colwise",
|
| 184 |
+
"blocks.*.mlp.ff_out": "rowwise",
|
| 185 |
+
}
|
| 186 |
+
base_model_pp_plan = {
|
| 187 |
+
"wte": (["input_ids"], ["inputs_embeds"]),
|
| 188 |
+
"blocks": (["hidden_states", "attention_mask"], ["hidden_states"]),
|
| 189 |
+
"ln_f": (["hidden_states"], ["hidden_states"]),
|
| 190 |
+
}
|
| 191 |
+
|
| 192 |
+
def __init__(
|
| 193 |
+
self,
|
| 194 |
+
hidden_size: int = 3584,
|
| 195 |
+
num_attention_heads: int = 28,
|
| 196 |
+
num_key_value_heads: Optional[int] = 4,
|
| 197 |
+
head_dim: int = 128,
|
| 198 |
+
vocab_size: int = 152064,
|
| 199 |
+
additional_vocab_size: int = 128,
|
| 200 |
+
qkv_bias: bool = True,
|
| 201 |
+
num_hidden_layers: int = 48,
|
| 202 |
+
intermediate_size: int = 18944,
|
| 203 |
+
hidden_act: str = "silu",
|
| 204 |
+
embedding_dropout: float=0.0,
|
| 205 |
+
attention_dropout: float=0.0,
|
| 206 |
+
residual_dropout: float = 0.0,
|
| 207 |
+
max_position_embeddings: int = 4096,
|
| 208 |
+
rope_theta: float = 1000000.0,
|
| 209 |
+
rope_scaling: dict[str, Any] = None,
|
| 210 |
+
rope_scaling_layers: Optional[list[int]] = None,
|
| 211 |
+
use_qk_norm: bool = False,
|
| 212 |
+
qk_norm_type: str = "olmo",
|
| 213 |
+
layer_norm_eps: int = 1e-6,
|
| 214 |
+
norm_after: bool = False,
|
| 215 |
+
initializer_range: float = 0.02,
|
| 216 |
+
use_cache=True,
|
| 217 |
+
tie_word_embeddings=False,
|
| 218 |
+
attn_implementation: str = "eager",
|
| 219 |
+
**kwargs,
|
| 220 |
+
):
|
| 221 |
+
self.attn_implementation = attn_implementation
|
| 222 |
+
super().__init__(
|
| 223 |
+
tie_word_embeddings=tie_word_embeddings,
|
| 224 |
+
attn_implementation=attn_implementation,
|
| 225 |
+
**kwargs
|
| 226 |
+
)
|
| 227 |
+
self.hidden_size = hidden_size
|
| 228 |
+
self.num_attention_heads = num_attention_heads
|
| 229 |
+
if num_key_value_heads is None:
|
| 230 |
+
num_key_value_heads = num_attention_heads
|
| 231 |
+
self.num_key_value_heads = num_key_value_heads
|
| 232 |
+
self.head_dim = head_dim
|
| 233 |
+
self.vocab_size = vocab_size
|
| 234 |
+
self.additional_vocab_size = additional_vocab_size
|
| 235 |
+
self.qkv_bias = qkv_bias
|
| 236 |
+
self.num_hidden_layers = num_hidden_layers
|
| 237 |
+
self.intermediate_size = intermediate_size
|
| 238 |
+
self.hidden_act = hidden_act
|
| 239 |
+
self.embedding_dropout = embedding_dropout
|
| 240 |
+
self.attention_dropout = attention_dropout
|
| 241 |
+
self.residual_dropout = residual_dropout
|
| 242 |
+
self.max_position_embeddings = max_position_embeddings
|
| 243 |
+
self.rope_theta = rope_theta
|
| 244 |
+
self.rope_scaling = rope_scaling
|
| 245 |
+
self.rope_scaling_layers = rope_scaling_layers
|
| 246 |
+
self.use_qk_norm = use_qk_norm
|
| 247 |
+
self.qk_norm_type = qk_norm_type
|
| 248 |
+
self.layer_norm_eps = layer_norm_eps
|
| 249 |
+
self.norm_after = norm_after
|
| 250 |
+
self.initializer_range = initializer_range
|
| 251 |
+
self.use_cache = use_cache
|
| 252 |
+
|
| 253 |
+
# Validate the correctness of rotary position embeddings parameters
|
| 254 |
+
rope_config_validation(self)
|
| 255 |
+
|
| 256 |
+
|
| 257 |
+
class Molmo2Config(PretrainedConfig):
|
| 258 |
+
r"""
|
| 259 |
+
This is the configuration class to store the configuration of a [`Molmo2ForConditionalGeneration`].
|
| 260 |
+
It is used to instantiate an Molmo2 model according to the specified arguments, defining the model architecture.
|
| 261 |
+
|
| 262 |
+
Example:
|
| 263 |
+
|
| 264 |
+
```python
|
| 265 |
+
>>> from transformers import Molmo2Config, Molmo2VitConfig, Molmo2AdapterConfig, Molmo2TextConfig
|
| 266 |
+
|
| 267 |
+
>>> # Initializing a Molmo2VitConfig
|
| 268 |
+
>>> vit_config = Molmo2VitConfig()
|
| 269 |
+
|
| 270 |
+
>>> # Initializing a Molmo2AdapterConfig
|
| 271 |
+
>>> adapter_config = Molmo2AdapterConfig()
|
| 272 |
+
|
| 273 |
+
>>> # Initializing a Molmo2TextConfig
|
| 274 |
+
>>> text_config = Molmo2TextConfig()
|
| 275 |
+
|
| 276 |
+
>>> # Initializing a Molmo2Config
|
| 277 |
+
>>> configuration = Molmo2Config(
|
| 278 |
+
>>> vit_config=vit_config,
|
| 279 |
+
>>> adapter_config=adapter_config,
|
| 280 |
+
>>> text_config=text_config,
|
| 281 |
+
>>> image_start_token_id=151936,
|
| 282 |
+
>>> image_end_token_id=151937,
|
| 283 |
+
>>> image_patch_id=151938,
|
| 284 |
+
>>> image_col_id=151939,
|
| 285 |
+
>>> low_res_image_start_token_id=151940,
|
| 286 |
+
>>> image_low_res_id=151942,
|
| 287 |
+
>>> frame_start_token_id=151943,
|
| 288 |
+
>>> frame_end_token_id=151944,
|
| 289 |
+
>>> )
|
| 290 |
+
|
| 291 |
+
>>> # Initializing a model
|
| 292 |
+
>>> model = Molmo2ForConditionalGeneration(configuration)
|
| 293 |
+
|
| 294 |
+
>>> # Accessing the model configuration
|
| 295 |
+
>>> configuration = model.config
|
| 296 |
+
```"""
|
| 297 |
+
|
| 298 |
+
model_type = "molmo2"
|
| 299 |
+
sub_configs = {
|
| 300 |
+
"text_config": Molmo2TextConfig,
|
| 301 |
+
"vit_config": Molmo2VitConfig,
|
| 302 |
+
"adapter_config": Molmo2AdapterConfig,
|
| 303 |
+
}
|
| 304 |
+
|
| 305 |
+
def __init__(
|
| 306 |
+
self,
|
| 307 |
+
vit_config: Molmo2VitConfig = None,
|
| 308 |
+
adapter_config: Molmo2AdapterConfig = None,
|
| 309 |
+
text_config: Molmo2TextConfig = None,
|
| 310 |
+
image_start_token_id: int = None,
|
| 311 |
+
low_res_image_start_token_id: int = None,
|
| 312 |
+
image_end_token_id: int = None,
|
| 313 |
+
image_low_res_id: int = None,
|
| 314 |
+
image_patch_id: int = None,
|
| 315 |
+
image_col_id: int = None,
|
| 316 |
+
frame_start_token_id: int = None,
|
| 317 |
+
frame_end_token_id: int = None,
|
| 318 |
+
use_frame_special_tokens: bool = True,
|
| 319 |
+
initializer_range: float = 0.02,
|
| 320 |
+
**kwargs,
|
| 321 |
+
):
|
| 322 |
+
super().__init__(**kwargs)
|
| 323 |
+
if vit_config is None:
|
| 324 |
+
self.vit_config = Molmo2VitConfig()
|
| 325 |
+
elif isinstance(vit_config, dict):
|
| 326 |
+
self.vit_config = Molmo2VitConfig(**vit_config)
|
| 327 |
+
else:
|
| 328 |
+
self.vit_config = vit_config
|
| 329 |
+
if adapter_config is None:
|
| 330 |
+
self.adapter_config = Molmo2AdapterConfig()
|
| 331 |
+
elif isinstance(adapter_config, dict):
|
| 332 |
+
self.adapter_config = Molmo2AdapterConfig(**adapter_config)
|
| 333 |
+
else:
|
| 334 |
+
self.adapter_config = adapter_config
|
| 335 |
+
if text_config is None:
|
| 336 |
+
self.text_config = Molmo2TextConfig()
|
| 337 |
+
elif isinstance(text_config, dict):
|
| 338 |
+
self.text_config = Molmo2TextConfig(**text_config)
|
| 339 |
+
else:
|
| 340 |
+
self.text_config = text_config
|
| 341 |
+
self.image_start_token_id = image_start_token_id
|
| 342 |
+
self.low_res_image_start_token_id = low_res_image_start_token_id
|
| 343 |
+
self.image_end_token_id = image_end_token_id
|
| 344 |
+
self.image_low_res_id = image_low_res_id
|
| 345 |
+
self.image_high_res_id = image_patch_id
|
| 346 |
+
self.image_patch_id = image_patch_id
|
| 347 |
+
self.image_col_id = image_col_id
|
| 348 |
+
self.frame_start_token_id = frame_start_token_id
|
| 349 |
+
self.frame_end_token_id = frame_end_token_id
|
| 350 |
+
self.use_frame_special_tokens = use_frame_special_tokens
|
| 351 |
+
self.initializer_range = initializer_range
|
| 352 |
+
|
| 353 |
+
@property
|
| 354 |
+
def image_num_patch(self):
|
| 355 |
+
assert self.vit_config is not None
|
| 356 |
+
return self.vit_config.image_num_patch
|
| 357 |
+
|
| 358 |
+
@property
|
| 359 |
+
def num_attention_heads(self):
|
| 360 |
+
return self.text_config.num_attention_heads
|
| 361 |
+
|
| 362 |
+
@property
|
| 363 |
+
def num_key_value_heads(self):
|
| 364 |
+
return self.text_config.num_key_value_heads
|
| 365 |
+
|
| 366 |
+
@property
|
| 367 |
+
def head_dim(self):
|
| 368 |
+
return self.text_config.head_dim
|
| 369 |
+
|
| 370 |
+
@property
|
| 371 |
+
def num_hidden_layers(self):
|
| 372 |
+
return self.text_config.num_hidden_layers
|
| 373 |
+
|
| 374 |
+
@property
|
| 375 |
+
def hidden_size(self):
|
| 376 |
+
return self.text_config.hidden_size
|
| 377 |
+
|
| 378 |
+
@property
|
| 379 |
+
def vocab_size(self):
|
| 380 |
+
return self.text_config.vocab_size
|
| 381 |
+
|
| 382 |
+
@property
|
| 383 |
+
def max_position_embeddings(self):
|
| 384 |
+
return self.text_config.max_position_embeddings
|
| 385 |
+
|
| 386 |
+
|
| 387 |
+
Molmo2VitConfig.register_for_auto_class()
|
| 388 |
+
Molmo2AdapterConfig.register_for_auto_class()
|
| 389 |
+
Molmo2TextConfig.register_for_auto_class()
|
| 390 |
+
Molmo2Config.register_for_auto_class()
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 151645,
|
| 3 |
+
"eos_token_id": 151645,
|
| 4 |
+
"pad_token_id": 151643,
|
| 5 |
+
"transformers_version": "4.47.0"
|
| 6 |
+
}
|
image_processing_molmo2.py
ADDED
|
@@ -0,0 +1,515 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""Image processor class for Molmo2"""
|
| 2 |
+
from typing import Optional, Union
|
| 3 |
+
import numpy as np
|
| 4 |
+
import einops
|
| 5 |
+
import torch
|
| 6 |
+
import torchvision.transforms
|
| 7 |
+
|
| 8 |
+
from transformers.image_utils import (
|
| 9 |
+
IMAGENET_STANDARD_MEAN,
|
| 10 |
+
IMAGENET_STANDARD_STD,
|
| 11 |
+
ImageInput,
|
| 12 |
+
PILImageResampling,
|
| 13 |
+
make_flat_list_of_images,
|
| 14 |
+
valid_images,
|
| 15 |
+
to_numpy_array,
|
| 16 |
+
)
|
| 17 |
+
from transformers.image_transforms import convert_to_rgb
|
| 18 |
+
from transformers.processing_utils import ImagesKwargs
|
| 19 |
+
from transformers.image_processing_utils import BaseImageProcessor, get_size_dict
|
| 20 |
+
from transformers.utils import logging
|
| 21 |
+
from transformers.feature_extraction_utils import BatchFeature
|
| 22 |
+
from transformers.utils import TensorType, logging
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
logger = logging.get_logger(__name__)
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
def normalize_image(
|
| 29 |
+
image: np.ndarray,
|
| 30 |
+
image_mean: list[float],
|
| 31 |
+
image_std: list[float],
|
| 32 |
+
) -> np.ndarray:
|
| 33 |
+
image -= np.array(image_mean, dtype=np.float32)[None, None, :]
|
| 34 |
+
image /= np.array(image_std, dtype=np.float32)[None, None, :]
|
| 35 |
+
return image
|
| 36 |
+
|
| 37 |
+
|
| 38 |
+
def resize_image(
|
| 39 |
+
image: np.ndarray,
|
| 40 |
+
desired_output_size: list[int],
|
| 41 |
+
resample: PILImageResampling,
|
| 42 |
+
) -> np.ndarray:
|
| 43 |
+
image = torch.permute(torch.from_numpy(image), [2, 0, 1])
|
| 44 |
+
dtype = image.dtype
|
| 45 |
+
if torch.is_floating_point(image):
|
| 46 |
+
in_min = 0.0
|
| 47 |
+
in_max = 1.0
|
| 48 |
+
resized = torchvision.transforms.Resize(
|
| 49 |
+
desired_output_size,
|
| 50 |
+
resample,
|
| 51 |
+
antialias=False,
|
| 52 |
+
)(image)
|
| 53 |
+
resized = torch.clip(resized, 0.0, 1.0).to(dtype)
|
| 54 |
+
else:
|
| 55 |
+
assert image.dtype == torch.uint8, "SigLIP expects float images or uint8 images, but got {}".format(image.dtype)
|
| 56 |
+
in_min = 0.0
|
| 57 |
+
in_max = 255.0
|
| 58 |
+
resized = torchvision.transforms.Resize(
|
| 59 |
+
desired_output_size,
|
| 60 |
+
resample,
|
| 61 |
+
antialias=False,
|
| 62 |
+
)(image)
|
| 63 |
+
resized = torch.clip(resized, 0, 255).to(dtype)
|
| 64 |
+
|
| 65 |
+
resized = resized.to(torch.float32)
|
| 66 |
+
resized = (resized - in_min) / (in_max - in_min)
|
| 67 |
+
|
| 68 |
+
resized = torch.permute(resized, [1, 2, 0]).numpy()
|
| 69 |
+
|
| 70 |
+
return resized
|
| 71 |
+
|
| 72 |
+
|
| 73 |
+
def select_tiling(h, w, patch_size, max_num_crops):
|
| 74 |
+
"""Divide in image of size [w, h] in up to max_num_patches of size patch_size"""
|
| 75 |
+
original_size = np.stack([h, w]) # [1, 2]
|
| 76 |
+
original_res = h * w
|
| 77 |
+
tilings = []
|
| 78 |
+
for i in range(1, max_num_crops + 1):
|
| 79 |
+
for j in range(1, max_num_crops + 1):
|
| 80 |
+
if i*j <= max_num_crops:
|
| 81 |
+
tilings.append((i, j))
|
| 82 |
+
# sort so argmin and argmax favour smaller tilings in the event of a tie
|
| 83 |
+
tilings.sort(key=lambda x: (x[0]*x[1], x[0]))
|
| 84 |
+
candidate_tilings = np.array(tilings, dtype=np.int32) # [n_resolutions, 2]
|
| 85 |
+
candidate_resolutions = candidate_tilings * patch_size # [n_resolutions, 2]
|
| 86 |
+
|
| 87 |
+
# How much we would need to scale the image to fit exactly in each tiling
|
| 88 |
+
original_size = np.stack([h, w], dtype=np.float32) # [1, 2]
|
| 89 |
+
|
| 90 |
+
# The original size can be zero in rare cases if the image is smaller than the margin
|
| 91 |
+
# In those cases letting the scale become infinite means the tiling is based on the
|
| 92 |
+
# other side, or falls back to the smallest tiling
|
| 93 |
+
with np.errstate(divide='ignore'):
|
| 94 |
+
required_scale_d = candidate_resolutions.astype(np.float32) / original_size,
|
| 95 |
+
required_scale = np.min(required_scale_d, axis=-1, keepdims=True) # [n_resolutions, 1]
|
| 96 |
+
if np.all(required_scale < 1):
|
| 97 |
+
# We are forced to downscale, so try to minimize the amount of downscaling
|
| 98 |
+
ix = np.argmax(required_scale)
|
| 99 |
+
else:
|
| 100 |
+
# Pick the resolution that required the least upscaling so that it most closely fits the image
|
| 101 |
+
required_scale = np.where(required_scale < 1.0, 10e9, required_scale)
|
| 102 |
+
ix = np.argmin(required_scale)
|
| 103 |
+
return candidate_tilings[ix]
|
| 104 |
+
|
| 105 |
+
|
| 106 |
+
def build_resized_image(
|
| 107 |
+
image: np.ndarray,
|
| 108 |
+
base_image_input_size: list[int],
|
| 109 |
+
resample: PILImageResampling,
|
| 110 |
+
image_mean: list[float],
|
| 111 |
+
image_std: list[float],
|
| 112 |
+
image_patch_size: int,
|
| 113 |
+
) -> tuple[np.ndarray, np.ndarray]:
|
| 114 |
+
resized = resize_image(
|
| 115 |
+
image, base_image_input_size, resample,
|
| 116 |
+
)
|
| 117 |
+
resized = normalize_image(resized, image_mean, image_std)
|
| 118 |
+
if len(resized.shape) == 3:
|
| 119 |
+
resized = np.expand_dims(resized, 0)
|
| 120 |
+
crop_patch_w = base_image_input_size[1] // image_patch_size
|
| 121 |
+
crop_patch_h = base_image_input_size[0] // image_patch_size
|
| 122 |
+
resize_idx = np.arange(crop_patch_w*crop_patch_h).reshape([crop_patch_h, crop_patch_w])
|
| 123 |
+
return resized, resize_idx
|
| 124 |
+
|
| 125 |
+
|
| 126 |
+
def build_overlapping_crops(
|
| 127 |
+
image: np.ndarray,
|
| 128 |
+
max_crops: int,
|
| 129 |
+
overlap_margins: list[int],
|
| 130 |
+
base_image_input_size: list[int],
|
| 131 |
+
resample: PILImageResampling,
|
| 132 |
+
image_mean: list[float],
|
| 133 |
+
image_std: list[float],
|
| 134 |
+
image_patch_size: int,
|
| 135 |
+
) -> tuple[np.ndarray, np.ndarray]:
|
| 136 |
+
"""Decompose an image into a set of overlapping crops
|
| 137 |
+
|
| 138 |
+
:return crop_arr: [n_crops, h, w, 3] The crops
|
| 139 |
+
:return patch_idx: [overlap_patch_h, overlap_patch_w] For each patch in the resized image
|
| 140 |
+
the crops were extracted from, what patch in `crop_arr` it corresponds to
|
| 141 |
+
"""
|
| 142 |
+
original_image_h, original_image_w = image.shape[:2]
|
| 143 |
+
crop_size = base_image_input_size[0]
|
| 144 |
+
assert base_image_input_size[0] == base_image_input_size[1]
|
| 145 |
+
|
| 146 |
+
left_margin, right_margin = overlap_margins
|
| 147 |
+
total_margin_pixels = image_patch_size * (right_margin + left_margin) # pixels removed per dim
|
| 148 |
+
crop_patches = base_image_input_size[0] // image_patch_size # patches per crop dim
|
| 149 |
+
crop_window_patches = crop_patches - (right_margin + left_margin) # usable patches
|
| 150 |
+
crop_window_size = crop_window_patches * image_patch_size
|
| 151 |
+
crop_patch_w = base_image_input_size[1] // image_patch_size
|
| 152 |
+
crop_patch_h = base_image_input_size[0] // image_patch_size
|
| 153 |
+
original_image_h, original_image_w = image.shape[:2]
|
| 154 |
+
crop_size = base_image_input_size[0]
|
| 155 |
+
|
| 156 |
+
# Decide how to tile the image, to account for the overlap margins we compute the tiling
|
| 157 |
+
# as if we had an image without the margins and were using a crop size without the margins
|
| 158 |
+
tiling = select_tiling(
|
| 159 |
+
max(original_image_h - total_margin_pixels, 1),
|
| 160 |
+
max(original_image_w - total_margin_pixels, 1),
|
| 161 |
+
crop_window_size,
|
| 162 |
+
max_crops,
|
| 163 |
+
)
|
| 164 |
+
|
| 165 |
+
src = resize_image(
|
| 166 |
+
image,
|
| 167 |
+
[tiling[0]*crop_window_size+total_margin_pixels, tiling[1]*crop_window_size+total_margin_pixels],
|
| 168 |
+
resample,
|
| 169 |
+
)
|
| 170 |
+
src = normalize_image(src, image_mean, image_std)
|
| 171 |
+
|
| 172 |
+
# Now we have to split the image into crops, and track what patches came from
|
| 173 |
+
# where in `patch_idx_arr`
|
| 174 |
+
n_crops = tiling[0] * tiling[1]
|
| 175 |
+
crop_arr = np.zeros([n_crops, crop_size, crop_size, 3], dtype=src.dtype)
|
| 176 |
+
patch_idx_arr = np.zeros([n_crops, crop_patch_h, crop_patch_w], dtype=np.int32)
|
| 177 |
+
on_crop = 0
|
| 178 |
+
for i in range(tiling[0]):
|
| 179 |
+
# Slide over `src` by `crop_window_size` steps, but extract crops of size `crops_size`
|
| 180 |
+
# which results in overlapping crop windows
|
| 181 |
+
y0 = i*crop_window_size
|
| 182 |
+
for j in range(tiling[1]):
|
| 183 |
+
x0 = j*crop_window_size
|
| 184 |
+
crop_arr[on_crop] = src[y0:y0+crop_size, x0:x0+crop_size]
|
| 185 |
+
patch_idx = np.arange(crop_patch_w*crop_patch_h).reshape(crop_patch_h, crop_patch_w)
|
| 186 |
+
patch_idx += on_crop * crop_patch_h * crop_patch_w
|
| 187 |
+
|
| 188 |
+
# Mask out idx that are in the overlap region
|
| 189 |
+
if i != 0:
|
| 190 |
+
patch_idx[:left_margin, :] = -1
|
| 191 |
+
if j != 0:
|
| 192 |
+
patch_idx[:, :left_margin] = -1
|
| 193 |
+
if i != tiling[0]-1:
|
| 194 |
+
patch_idx[-right_margin:, :] = -1
|
| 195 |
+
if j != tiling[1]-1:
|
| 196 |
+
patch_idx[:, -right_margin:] = -1
|
| 197 |
+
patch_idx_arr[on_crop] = patch_idx
|
| 198 |
+
on_crop += 1
|
| 199 |
+
|
| 200 |
+
# `patch_idx_arr` is ordered crop-by-crop, here we transpose `patch_idx_arr`
|
| 201 |
+
# so it is ordered left-to-right order
|
| 202 |
+
patch_idx_arr = np.reshape(
|
| 203 |
+
patch_idx_arr,
|
| 204 |
+
[tiling[0], tiling[1], crop_patch_h, crop_patch_w]
|
| 205 |
+
)
|
| 206 |
+
patch_idx_arr = np.transpose(patch_idx_arr, [0, 2, 1, 3])
|
| 207 |
+
patch_idx_arr = np.reshape(patch_idx_arr, [-1])
|
| 208 |
+
|
| 209 |
+
# Now get the parts not in the overlap region, so it should map each patch in `src`
|
| 210 |
+
# to the correct patch it should come from in `crop_arr`
|
| 211 |
+
patch_idx_arr = patch_idx_arr[patch_idx_arr >= 0].reshape(
|
| 212 |
+
src.shape[0]//image_patch_size,
|
| 213 |
+
src.shape[1]//image_patch_size,
|
| 214 |
+
)
|
| 215 |
+
return crop_arr, patch_idx_arr
|
| 216 |
+
|
| 217 |
+
|
| 218 |
+
def batch_pixels_to_patches(array: np.ndarray, patch_size: int) -> np.ndarray:
|
| 219 |
+
"""Reshape images of [n_images, h, w, 3] -> [n_images, n_patches, pixels_per_patch]"""
|
| 220 |
+
if len(array.shape) == 3:
|
| 221 |
+
n_crops, h, w = array.shape
|
| 222 |
+
h_patches = h//patch_size
|
| 223 |
+
w_patches = w//patch_size
|
| 224 |
+
array = np.reshape(array, [n_crops, h_patches, patch_size, w_patches, patch_size])
|
| 225 |
+
array = np.transpose(array, [0, 1, 3, 2, 4])
|
| 226 |
+
array = np.reshape(array, [n_crops, h_patches*w_patches, patch_size*patch_size])
|
| 227 |
+
return array
|
| 228 |
+
else:
|
| 229 |
+
n_crops, h, w, c = array.shape
|
| 230 |
+
h_patches = h//patch_size
|
| 231 |
+
w_patches = w//patch_size
|
| 232 |
+
array = np.reshape(array, [n_crops, h_patches, patch_size, w_patches, patch_size, c])
|
| 233 |
+
array = np.transpose(array, [0, 1, 3, 2, 4, 5])
|
| 234 |
+
array = np.reshape(array, [n_crops, h_patches*w_patches, patch_size*patch_size*c])
|
| 235 |
+
return array
|
| 236 |
+
|
| 237 |
+
|
| 238 |
+
def arange_for_pooling(
|
| 239 |
+
idx_arr: np.ndarray,
|
| 240 |
+
pool_h: int,
|
| 241 |
+
pool_w: int,
|
| 242 |
+
) -> np.ndarray:
|
| 243 |
+
h_pad = pool_h * ((idx_arr.shape[0] + pool_h - 1) // pool_h) - idx_arr.shape[0]
|
| 244 |
+
w_pad = pool_w * ((idx_arr.shape[1] + pool_w - 1) // pool_w) - idx_arr.shape[1]
|
| 245 |
+
idx_arr = np.pad(idx_arr, [[h_pad//2, (h_pad+1)//2], [w_pad//2, (w_pad+1)//2]],
|
| 246 |
+
mode='constant',constant_values=-1)
|
| 247 |
+
return einops.rearrange(
|
| 248 |
+
idx_arr, "(h dh) (w dw) -> h w (dh dw)", dh=pool_h, dw=pool_w)
|
| 249 |
+
|
| 250 |
+
|
| 251 |
+
def image_to_patches_and_grids(
|
| 252 |
+
image: np.ndarray,
|
| 253 |
+
max_crops: int,
|
| 254 |
+
overlap_margins: list[int],
|
| 255 |
+
base_image_input_size: list[int],
|
| 256 |
+
resample: PILImageResampling,
|
| 257 |
+
image_mean: list[float],
|
| 258 |
+
image_std: list[float],
|
| 259 |
+
image_patch_size: int,
|
| 260 |
+
image_pooling_w: int,
|
| 261 |
+
image_pooling_h: int,
|
| 262 |
+
) -> tuple[np.ndarray, np.ndarray, np.ndarray]:
|
| 263 |
+
"""
|
| 264 |
+
:return image_grids, the shape of each (low-res, high-res) image after pooling
|
| 265 |
+
:return crops, the image crops to processes with the ViT
|
| 266 |
+
:return pooled_patch_idx, for each patch_id tokens in `image_tokens`, the indices of the
|
| 267 |
+
patches in `crops` to pool for that token, masked with -1
|
| 268 |
+
"""
|
| 269 |
+
if isinstance(base_image_input_size, int):
|
| 270 |
+
base_image_input_size = (base_image_input_size, base_image_input_size)
|
| 271 |
+
|
| 272 |
+
base_image_input_d = image_patch_size
|
| 273 |
+
pooling_w = image_pooling_w
|
| 274 |
+
pooling_h = image_pooling_h
|
| 275 |
+
crop_patch_w = base_image_input_size[1] // base_image_input_d
|
| 276 |
+
crop_patch_h = base_image_input_size[0] // base_image_input_d
|
| 277 |
+
|
| 278 |
+
crop_arr, patch_idx_arr = build_overlapping_crops(
|
| 279 |
+
image,
|
| 280 |
+
max_crops,
|
| 281 |
+
overlap_margins,
|
| 282 |
+
base_image_input_size,
|
| 283 |
+
resample,
|
| 284 |
+
image_mean,
|
| 285 |
+
image_std,
|
| 286 |
+
image_patch_size,
|
| 287 |
+
)
|
| 288 |
+
pooling_idx = arange_for_pooling(patch_idx_arr, pooling_h, pooling_w)
|
| 289 |
+
h, w = pooling_idx.shape[:2]
|
| 290 |
+
pooling_idx = pooling_idx.reshape([-1, pooling_h*pooling_w])
|
| 291 |
+
|
| 292 |
+
# Finally do the same for the global image
|
| 293 |
+
resized, resize_idx = build_resized_image(
|
| 294 |
+
image,
|
| 295 |
+
base_image_input_size,
|
| 296 |
+
resample,
|
| 297 |
+
image_mean,
|
| 298 |
+
image_std,
|
| 299 |
+
image_patch_size,
|
| 300 |
+
)
|
| 301 |
+
crop_arr = np.concatenate([resized, crop_arr], 0)
|
| 302 |
+
|
| 303 |
+
resize_idx = arange_for_pooling(resize_idx, pooling_h, pooling_w)
|
| 304 |
+
resized_h, resized_w = resize_idx.shape[:2]
|
| 305 |
+
resize_idx = resize_idx.reshape([-1, pooling_h*pooling_w])
|
| 306 |
+
|
| 307 |
+
# Global image goes first, so the order of patches in previous crops gets increased
|
| 308 |
+
pooling_idx = np.where(
|
| 309 |
+
pooling_idx >= 0,
|
| 310 |
+
pooling_idx + crop_patch_h*crop_patch_w,
|
| 311 |
+
-1
|
| 312 |
+
)
|
| 313 |
+
pooling_idx = np.concatenate([resize_idx, pooling_idx])
|
| 314 |
+
image_grid = [np.array([resized_h, resized_w, h, w])]
|
| 315 |
+
|
| 316 |
+
return (
|
| 317 |
+
np.stack(image_grid, 0),
|
| 318 |
+
batch_pixels_to_patches(crop_arr, image_patch_size),
|
| 319 |
+
pooling_idx
|
| 320 |
+
)
|
| 321 |
+
|
| 322 |
+
|
| 323 |
+
class Molmo2ImagesKwargs(ImagesKwargs, total=False):
|
| 324 |
+
max_crops: Optional[int]
|
| 325 |
+
overlap_margins: Optional[list[int]]
|
| 326 |
+
patch_size: Optional[int]
|
| 327 |
+
pooling_size: Optional[list[int]]
|
| 328 |
+
|
| 329 |
+
|
| 330 |
+
class Molmo2ImageProcessor(BaseImageProcessor):
|
| 331 |
+
r"""
|
| 332 |
+
Constructs a Molmo2 image processor that preprocesses images for the model.
|
| 333 |
+
|
| 334 |
+
Args:
|
| 335 |
+
size (`dict[str, int]` *optional*, defaults to `{"height": 378, "width": 378}`):
|
| 336 |
+
Size of the image after resizing.
|
| 337 |
+
resample (`PILImageResampling`, *optional*, defaults to `Resampling.BILINEAR`):
|
| 338 |
+
Resampling filter to use when resizing the image.
|
| 339 |
+
image_mean (`float` or `list[float]`, *optional*, defaults to `[0.5, 0.5, 0.5]`):
|
| 340 |
+
Mean to use if normalizing the image. This is a float or list of floats for each channel in the image.
|
| 341 |
+
image_std (`float` or `list[float]`, *optional*, defaults to `[0.5, 0.5, 0.5]`):
|
| 342 |
+
Standard deviation to use if normalizing the image. This is a float or list of floats for each channel in the image.
|
| 343 |
+
do_convert_rgb (`bool`, *optional*, defaults to `True`):
|
| 344 |
+
Whether to convert the image to RGB.
|
| 345 |
+
max_crops (`int`, *optional*, defaults to `8`):
|
| 346 |
+
Maximum number of crops to use per image.
|
| 347 |
+
overlap_margins (`list[int]`, *optional*, defaults to `[4, 4]`):
|
| 348 |
+
Overlap margins to use.
|
| 349 |
+
patch_size (`int`, *optional*, defaults to 14):
|
| 350 |
+
The spatial patch size of the vision encoder.
|
| 351 |
+
pooling_size (`list[int]`, *optional*, defaults to `[2, 2]`):
|
| 352 |
+
The pooling size of the vision adapter.
|
| 353 |
+
"""
|
| 354 |
+
|
| 355 |
+
model_input_names = ["pixel_values", "image_token_pooling", "image_grids", "image_num_crops"]
|
| 356 |
+
|
| 357 |
+
def __init__(
|
| 358 |
+
self,
|
| 359 |
+
size: Optional[dict[str, int]] = None,
|
| 360 |
+
resample: PILImageResampling = PILImageResampling.BILINEAR,
|
| 361 |
+
image_mean: Optional[Union[float, list[float]]] = None,
|
| 362 |
+
image_std: Optional[Union[float, list[float]]] = None,
|
| 363 |
+
do_convert_rgb: bool = True,
|
| 364 |
+
max_crops: int = 8,
|
| 365 |
+
overlap_margins: list[int] = [4, 4],
|
| 366 |
+
patch_size: int = 14,
|
| 367 |
+
pooling_size: list[int] = [2, 2],
|
| 368 |
+
**kwargs,
|
| 369 |
+
) -> None:
|
| 370 |
+
super().__init__(**kwargs)
|
| 371 |
+
size = size if size is not None else {"height": 378, "width": 378}
|
| 372 |
+
size = get_size_dict(size, default_to_square=True)
|
| 373 |
+
self.size = size
|
| 374 |
+
|
| 375 |
+
self.resample = resample
|
| 376 |
+
self.image_mean = image_mean if image_mean is not None else IMAGENET_STANDARD_MEAN
|
| 377 |
+
self.image_std = image_std if image_std is not None else IMAGENET_STANDARD_STD
|
| 378 |
+
self.do_convert_rgb = do_convert_rgb
|
| 379 |
+
|
| 380 |
+
self.max_crops = max_crops
|
| 381 |
+
self.overlap_margins = overlap_margins
|
| 382 |
+
self.patch_size = patch_size
|
| 383 |
+
self.pooling_size = pooling_size
|
| 384 |
+
|
| 385 |
+
def preprocess(
|
| 386 |
+
self,
|
| 387 |
+
images: ImageInput,
|
| 388 |
+
size: Optional[dict[str, int]] = None,
|
| 389 |
+
resample: Optional[PILImageResampling] = None,
|
| 390 |
+
image_mean: Optional[Union[float, list[float]]] = None,
|
| 391 |
+
image_std: Optional[Union[float, list[float]]] = None,
|
| 392 |
+
do_convert_rgb: Optional[bool] = None,
|
| 393 |
+
max_crops: Optional[int] = None,
|
| 394 |
+
overlap_margins: Optional[list[int]] = None,
|
| 395 |
+
patch_size: Optional[int] = None,
|
| 396 |
+
pooling_size: Optional[list[int]] = None,
|
| 397 |
+
return_tensors: Optional[Union[str, TensorType]] = None,
|
| 398 |
+
**kwargs,
|
| 399 |
+
) -> BatchFeature:
|
| 400 |
+
"""
|
| 401 |
+
Args:
|
| 402 |
+
images (`ImageInput`):
|
| 403 |
+
Image to preprocess.
|
| 404 |
+
size (`dict[str, int]`, *optional*, defaults to `self.size`):
|
| 405 |
+
Size of the image after resizing.
|
| 406 |
+
resample (`PILImageResampling`, *optional*, defaults to `self.resample`):
|
| 407 |
+
Resampling filter to use when resizing the image. This can be one of the enum `PILImageResampling`. Only
|
| 408 |
+
has an effect if `do_resize` is set to `True`.
|
| 409 |
+
image_mean (`float` or `list[float]`, *optional*, defaults to `self.image_mean`):
|
| 410 |
+
Image mean to use for normalization. Only has an effect if `do_normalize` is set to `True`.
|
| 411 |
+
image_std (`float` or `list[float]`, *optional*, defaults to `self.image_std`):
|
| 412 |
+
Image standard deviation to use for normalization. Only has an effect if `do_normalize` is set to
|
| 413 |
+
`True`.
|
| 414 |
+
do_convert_rgb (`bool`, *optional*, defaults to `self.do_convert_rgb`):
|
| 415 |
+
Whether to convert the image to RGB.
|
| 416 |
+
max_crops (`int`, *optional*, defaults to `self.max_crops`):
|
| 417 |
+
Maximum number of crops to use per image.
|
| 418 |
+
overlap_margins (`list[int]`, *optional*, defaults to `self.overlap_margins`):
|
| 419 |
+
Overlap margins to use.
|
| 420 |
+
patch_size (`int`, *optional*, defaults to `self.patch_size`):
|
| 421 |
+
The spatial patch size of the vision encoder.
|
| 422 |
+
pooling_size (`list[int]`, *optional*, defaults to `self.pooling_size`):
|
| 423 |
+
The pooling size of the vision adapter.
|
| 424 |
+
return_tensors (`str` or `TensorType`, *optional*):
|
| 425 |
+
The type of tensors to return. Can be one of:
|
| 426 |
+
- Unset: Return a list of `np.ndarray`.
|
| 427 |
+
- `TensorType.TENSORFLOW` or `'tf'`: Return a batch of type `tf.Tensor`.
|
| 428 |
+
- `TensorType.PYTORCH` or `'pt'`: Return a batch of type `torch.Tensor`.
|
| 429 |
+
- `TensorType.NUMPY` or `'np'`: Return a batch of type `np.ndarray`.
|
| 430 |
+
- `TensorType.JAX` or `'jax'`: Return a batch of type `jax.numpy.ndarray`.
|
| 431 |
+
|
| 432 |
+
Returns:
|
| 433 |
+
A `BatchFeature` containing the following keys:
|
| 434 |
+
- `pixel_values`: The preprocessed images.
|
| 435 |
+
- `image_token_pooling`: The indices of the patches in `crops` to pool for each token in `image_tokens`.
|
| 436 |
+
- `image_grids`: The image grids.
|
| 437 |
+
- `image_num_crops`: The number of crops for each image.
|
| 438 |
+
"""
|
| 439 |
+
if size is not None:
|
| 440 |
+
if "height" not in size or "width" not in size:
|
| 441 |
+
raise ValueError("size must contain 'height' and 'width' keys.")
|
| 442 |
+
else:
|
| 443 |
+
size = {**self.size}
|
| 444 |
+
|
| 445 |
+
base_image_input_size = [size["height"], size["width"]]
|
| 446 |
+
|
| 447 |
+
resample = resample or self.resample
|
| 448 |
+
image_mean = image_mean or self.image_mean
|
| 449 |
+
image_std = image_std or self.image_std
|
| 450 |
+
do_convert_rgb = do_convert_rgb or self.do_convert_rgb
|
| 451 |
+
|
| 452 |
+
max_crops = max_crops or self.max_crops
|
| 453 |
+
overlap_margins = overlap_margins or self.overlap_margins
|
| 454 |
+
patch_size = patch_size or self.patch_size
|
| 455 |
+
pooling_size = pooling_size or self.pooling_size
|
| 456 |
+
|
| 457 |
+
image_pooling_h, image_pooling_w = pooling_size
|
| 458 |
+
|
| 459 |
+
if images is not None:
|
| 460 |
+
images = self.fetch_images(images)
|
| 461 |
+
images = make_flat_list_of_images(images)
|
| 462 |
+
|
| 463 |
+
if images is not None and not valid_images(images):
|
| 464 |
+
raise ValueError(
|
| 465 |
+
"Invalid image type. Must be of type PIL.Image.Image, numpy.ndarray, "
|
| 466 |
+
"torch.Tensor, tf.Tensor or jax.ndarray."
|
| 467 |
+
)
|
| 468 |
+
|
| 469 |
+
if do_convert_rgb:
|
| 470 |
+
images = [convert_to_rgb(image) for image in images]
|
| 471 |
+
|
| 472 |
+
# All transformations expect numpy arrays.
|
| 473 |
+
images = [to_numpy_array(image) for image in images]
|
| 474 |
+
|
| 475 |
+
data = {}
|
| 476 |
+
if images is not None:
|
| 477 |
+
batch_grids = []
|
| 478 |
+
batch_crops = []
|
| 479 |
+
batch_pooled_patches_idx = []
|
| 480 |
+
batch_num_crops = []
|
| 481 |
+
|
| 482 |
+
for image in images:
|
| 483 |
+
image_grid, crops, pooled_idx = image_to_patches_and_grids(
|
| 484 |
+
image,
|
| 485 |
+
max_crops,
|
| 486 |
+
overlap_margins,
|
| 487 |
+
base_image_input_size,
|
| 488 |
+
resample,
|
| 489 |
+
image_mean,
|
| 490 |
+
image_std,
|
| 491 |
+
patch_size,
|
| 492 |
+
image_pooling_w,
|
| 493 |
+
image_pooling_h,
|
| 494 |
+
)
|
| 495 |
+
batch_grids.append(image_grid)
|
| 496 |
+
batch_crops.append(crops)
|
| 497 |
+
batch_pooled_patches_idx.append(pooled_idx)
|
| 498 |
+
batch_num_crops.append(crops.shape[0])
|
| 499 |
+
|
| 500 |
+
pixel_values = np.concatenate(batch_crops, 0)
|
| 501 |
+
image_token_pooling = np.concatenate(batch_pooled_patches_idx, 0)
|
| 502 |
+
image_grids = np.concatenate(batch_grids, 0)
|
| 503 |
+
image_num_crops = np.array(batch_num_crops)
|
| 504 |
+
|
| 505 |
+
data.update(
|
| 506 |
+
pixel_values=pixel_values,
|
| 507 |
+
image_token_pooling=image_token_pooling,
|
| 508 |
+
image_grids=image_grids,
|
| 509 |
+
image_num_crops=image_num_crops,
|
| 510 |
+
)
|
| 511 |
+
|
| 512 |
+
return BatchFeature(data, tensor_type=return_tensors)
|
| 513 |
+
|
| 514 |
+
|
| 515 |
+
Molmo2ImageProcessor.register_for_auto_class()
|
merges.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
model-00001-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bbabcf0e7c378850561acb90dc285baaf7c1c13612bb66529c936cc0f5ecfcf1
|
| 3 |
+
size 4891799000
|
model-00002-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:481e451be934617f94754e07ec9fde843a4f80e15f797c1fb6408ae027ce3755
|
| 3 |
+
size 4844690992
|
model-00003-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b6d9ef42149ba81545dff1df45da0354bcc3c66a6d7f70167dcd709c9f34a959
|
| 3 |
+
size 4844691024
|
model-00004-of-00004.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3889bb35a94ba82209943ffe2cc9f424e0b792c634b26590bd41d91089119e92
|
| 3 |
+
size 4822393416
|
model.safetensors.index.json
ADDED
|
@@ -0,0 +1,714 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"metadata": {
|
| 3 |
+
"total_parameters": 4850869200,
|
| 4 |
+
"total_size": 19403476800
|
| 5 |
+
},
|
| 6 |
+
"weight_map": {
|
| 7 |
+
"lm_head.weight": "model-00004-of-00004.safetensors",
|
| 8 |
+
"model.transformer.blocks.0.attn_norm.weight": "model-00001-of-00004.safetensors",
|
| 9 |
+
"model.transformer.blocks.0.ff_norm.weight": "model-00001-of-00004.safetensors",
|
| 10 |
+
"model.transformer.blocks.0.mlp.ff_out.weight": "model-00001-of-00004.safetensors",
|
| 11 |
+
"model.transformer.blocks.0.mlp.ff_proj.weight": "model-00001-of-00004.safetensors",
|
| 12 |
+
"model.transformer.blocks.0.self_attn.att_proj.weight": "model-00001-of-00004.safetensors",
|
| 13 |
+
"model.transformer.blocks.0.self_attn.attn_out.weight": "model-00001-of-00004.safetensors",
|
| 14 |
+
"model.transformer.blocks.0.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 15 |
+
"model.transformer.blocks.0.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 16 |
+
"model.transformer.blocks.1.attn_norm.weight": "model-00001-of-00004.safetensors",
|
| 17 |
+
"model.transformer.blocks.1.ff_norm.weight": "model-00001-of-00004.safetensors",
|
| 18 |
+
"model.transformer.blocks.1.mlp.ff_out.weight": "model-00001-of-00004.safetensors",
|
| 19 |
+
"model.transformer.blocks.1.mlp.ff_proj.weight": "model-00001-of-00004.safetensors",
|
| 20 |
+
"model.transformer.blocks.1.self_attn.att_proj.weight": "model-00001-of-00004.safetensors",
|
| 21 |
+
"model.transformer.blocks.1.self_attn.attn_out.weight": "model-00001-of-00004.safetensors",
|
| 22 |
+
"model.transformer.blocks.1.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 23 |
+
"model.transformer.blocks.1.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 24 |
+
"model.transformer.blocks.10.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 25 |
+
"model.transformer.blocks.10.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 26 |
+
"model.transformer.blocks.10.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 27 |
+
"model.transformer.blocks.10.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 28 |
+
"model.transformer.blocks.10.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 29 |
+
"model.transformer.blocks.10.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 30 |
+
"model.transformer.blocks.10.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 31 |
+
"model.transformer.blocks.10.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 32 |
+
"model.transformer.blocks.11.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 33 |
+
"model.transformer.blocks.11.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 34 |
+
"model.transformer.blocks.11.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 35 |
+
"model.transformer.blocks.11.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 36 |
+
"model.transformer.blocks.11.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 37 |
+
"model.transformer.blocks.11.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 38 |
+
"model.transformer.blocks.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 39 |
+
"model.transformer.blocks.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 40 |
+
"model.transformer.blocks.12.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 41 |
+
"model.transformer.blocks.12.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 42 |
+
"model.transformer.blocks.12.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 43 |
+
"model.transformer.blocks.12.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 44 |
+
"model.transformer.blocks.12.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 45 |
+
"model.transformer.blocks.12.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 46 |
+
"model.transformer.blocks.12.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 47 |
+
"model.transformer.blocks.12.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 48 |
+
"model.transformer.blocks.13.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 49 |
+
"model.transformer.blocks.13.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 50 |
+
"model.transformer.blocks.13.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 51 |
+
"model.transformer.blocks.13.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 52 |
+
"model.transformer.blocks.13.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 53 |
+
"model.transformer.blocks.13.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 54 |
+
"model.transformer.blocks.13.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 55 |
+
"model.transformer.blocks.13.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 56 |
+
"model.transformer.blocks.14.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 57 |
+
"model.transformer.blocks.14.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 58 |
+
"model.transformer.blocks.14.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 59 |
+
"model.transformer.blocks.14.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 60 |
+
"model.transformer.blocks.14.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 61 |
+
"model.transformer.blocks.14.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 62 |
+
"model.transformer.blocks.14.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 63 |
+
"model.transformer.blocks.14.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 64 |
+
"model.transformer.blocks.15.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 65 |
+
"model.transformer.blocks.15.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 66 |
+
"model.transformer.blocks.15.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 67 |
+
"model.transformer.blocks.15.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 68 |
+
"model.transformer.blocks.15.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 69 |
+
"model.transformer.blocks.15.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 70 |
+
"model.transformer.blocks.15.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 71 |
+
"model.transformer.blocks.15.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 72 |
+
"model.transformer.blocks.16.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 73 |
+
"model.transformer.blocks.16.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 74 |
+
"model.transformer.blocks.16.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 75 |
+
"model.transformer.blocks.16.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 76 |
+
"model.transformer.blocks.16.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 77 |
+
"model.transformer.blocks.16.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 78 |
+
"model.transformer.blocks.16.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 79 |
+
"model.transformer.blocks.16.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 80 |
+
"model.transformer.blocks.17.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 81 |
+
"model.transformer.blocks.17.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 82 |
+
"model.transformer.blocks.17.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 83 |
+
"model.transformer.blocks.17.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 84 |
+
"model.transformer.blocks.17.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 85 |
+
"model.transformer.blocks.17.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 86 |
+
"model.transformer.blocks.17.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 87 |
+
"model.transformer.blocks.17.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 88 |
+
"model.transformer.blocks.18.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 89 |
+
"model.transformer.blocks.18.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 90 |
+
"model.transformer.blocks.18.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 91 |
+
"model.transformer.blocks.18.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 92 |
+
"model.transformer.blocks.18.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 93 |
+
"model.transformer.blocks.18.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 94 |
+
"model.transformer.blocks.18.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 95 |
+
"model.transformer.blocks.18.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 96 |
+
"model.transformer.blocks.19.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 97 |
+
"model.transformer.blocks.19.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 98 |
+
"model.transformer.blocks.19.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 99 |
+
"model.transformer.blocks.19.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 100 |
+
"model.transformer.blocks.19.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 101 |
+
"model.transformer.blocks.19.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 102 |
+
"model.transformer.blocks.19.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 103 |
+
"model.transformer.blocks.19.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 104 |
+
"model.transformer.blocks.2.attn_norm.weight": "model-00001-of-00004.safetensors",
|
| 105 |
+
"model.transformer.blocks.2.ff_norm.weight": "model-00001-of-00004.safetensors",
|
| 106 |
+
"model.transformer.blocks.2.mlp.ff_out.weight": "model-00001-of-00004.safetensors",
|
| 107 |
+
"model.transformer.blocks.2.mlp.ff_proj.weight": "model-00001-of-00004.safetensors",
|
| 108 |
+
"model.transformer.blocks.2.self_attn.att_proj.weight": "model-00001-of-00004.safetensors",
|
| 109 |
+
"model.transformer.blocks.2.self_attn.attn_out.weight": "model-00001-of-00004.safetensors",
|
| 110 |
+
"model.transformer.blocks.2.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 111 |
+
"model.transformer.blocks.2.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 112 |
+
"model.transformer.blocks.20.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 113 |
+
"model.transformer.blocks.20.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 114 |
+
"model.transformer.blocks.20.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 115 |
+
"model.transformer.blocks.20.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 116 |
+
"model.transformer.blocks.20.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 117 |
+
"model.transformer.blocks.20.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 118 |
+
"model.transformer.blocks.20.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 119 |
+
"model.transformer.blocks.20.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 120 |
+
"model.transformer.blocks.21.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 121 |
+
"model.transformer.blocks.21.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 122 |
+
"model.transformer.blocks.21.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 123 |
+
"model.transformer.blocks.21.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 124 |
+
"model.transformer.blocks.21.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 125 |
+
"model.transformer.blocks.21.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 126 |
+
"model.transformer.blocks.21.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 127 |
+
"model.transformer.blocks.21.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 128 |
+
"model.transformer.blocks.22.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 129 |
+
"model.transformer.blocks.22.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 130 |
+
"model.transformer.blocks.22.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 131 |
+
"model.transformer.blocks.22.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 132 |
+
"model.transformer.blocks.22.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 133 |
+
"model.transformer.blocks.22.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 134 |
+
"model.transformer.blocks.22.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 135 |
+
"model.transformer.blocks.22.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 136 |
+
"model.transformer.blocks.23.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 137 |
+
"model.transformer.blocks.23.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 138 |
+
"model.transformer.blocks.23.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 139 |
+
"model.transformer.blocks.23.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 140 |
+
"model.transformer.blocks.23.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 141 |
+
"model.transformer.blocks.23.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 142 |
+
"model.transformer.blocks.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 143 |
+
"model.transformer.blocks.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 144 |
+
"model.transformer.blocks.24.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 145 |
+
"model.transformer.blocks.24.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 146 |
+
"model.transformer.blocks.24.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 147 |
+
"model.transformer.blocks.24.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 148 |
+
"model.transformer.blocks.24.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 149 |
+
"model.transformer.blocks.24.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 150 |
+
"model.transformer.blocks.24.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 151 |
+
"model.transformer.blocks.24.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 152 |
+
"model.transformer.blocks.25.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 153 |
+
"model.transformer.blocks.25.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 154 |
+
"model.transformer.blocks.25.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 155 |
+
"model.transformer.blocks.25.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 156 |
+
"model.transformer.blocks.25.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 157 |
+
"model.transformer.blocks.25.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 158 |
+
"model.transformer.blocks.25.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 159 |
+
"model.transformer.blocks.25.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 160 |
+
"model.transformer.blocks.26.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 161 |
+
"model.transformer.blocks.26.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 162 |
+
"model.transformer.blocks.26.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 163 |
+
"model.transformer.blocks.26.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 164 |
+
"model.transformer.blocks.26.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 165 |
+
"model.transformer.blocks.26.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 166 |
+
"model.transformer.blocks.26.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 167 |
+
"model.transformer.blocks.26.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 168 |
+
"model.transformer.blocks.27.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 169 |
+
"model.transformer.blocks.27.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 170 |
+
"model.transformer.blocks.27.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 171 |
+
"model.transformer.blocks.27.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 172 |
+
"model.transformer.blocks.27.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 173 |
+
"model.transformer.blocks.27.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 174 |
+
"model.transformer.blocks.27.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 175 |
+
"model.transformer.blocks.27.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 176 |
+
"model.transformer.blocks.28.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 177 |
+
"model.transformer.blocks.28.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 178 |
+
"model.transformer.blocks.28.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 179 |
+
"model.transformer.blocks.28.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 180 |
+
"model.transformer.blocks.28.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 181 |
+
"model.transformer.blocks.28.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 182 |
+
"model.transformer.blocks.28.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 183 |
+
"model.transformer.blocks.28.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 184 |
+
"model.transformer.blocks.29.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 185 |
+
"model.transformer.blocks.29.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 186 |
+
"model.transformer.blocks.29.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 187 |
+
"model.transformer.blocks.29.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 188 |
+
"model.transformer.blocks.29.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 189 |
+
"model.transformer.blocks.29.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 190 |
+
"model.transformer.blocks.29.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 191 |
+
"model.transformer.blocks.29.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 192 |
+
"model.transformer.blocks.3.attn_norm.weight": "model-00001-of-00004.safetensors",
|
| 193 |
+
"model.transformer.blocks.3.ff_norm.weight": "model-00001-of-00004.safetensors",
|
| 194 |
+
"model.transformer.blocks.3.mlp.ff_out.weight": "model-00001-of-00004.safetensors",
|
| 195 |
+
"model.transformer.blocks.3.mlp.ff_proj.weight": "model-00001-of-00004.safetensors",
|
| 196 |
+
"model.transformer.blocks.3.self_attn.att_proj.weight": "model-00001-of-00004.safetensors",
|
| 197 |
+
"model.transformer.blocks.3.self_attn.attn_out.weight": "model-00001-of-00004.safetensors",
|
| 198 |
+
"model.transformer.blocks.3.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 199 |
+
"model.transformer.blocks.3.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 200 |
+
"model.transformer.blocks.30.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 201 |
+
"model.transformer.blocks.30.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 202 |
+
"model.transformer.blocks.30.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 203 |
+
"model.transformer.blocks.30.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 204 |
+
"model.transformer.blocks.30.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 205 |
+
"model.transformer.blocks.30.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 206 |
+
"model.transformer.blocks.30.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 207 |
+
"model.transformer.blocks.30.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 208 |
+
"model.transformer.blocks.31.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 209 |
+
"model.transformer.blocks.31.ff_norm.weight": "model-00003-of-00004.safetensors",
|
| 210 |
+
"model.transformer.blocks.31.mlp.ff_out.weight": "model-00003-of-00004.safetensors",
|
| 211 |
+
"model.transformer.blocks.31.mlp.ff_proj.weight": "model-00003-of-00004.safetensors",
|
| 212 |
+
"model.transformer.blocks.31.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 213 |
+
"model.transformer.blocks.31.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 214 |
+
"model.transformer.blocks.31.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 215 |
+
"model.transformer.blocks.31.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 216 |
+
"model.transformer.blocks.32.attn_norm.weight": "model-00003-of-00004.safetensors",
|
| 217 |
+
"model.transformer.blocks.32.ff_norm.weight": "model-00004-of-00004.safetensors",
|
| 218 |
+
"model.transformer.blocks.32.mlp.ff_out.weight": "model-00004-of-00004.safetensors",
|
| 219 |
+
"model.transformer.blocks.32.mlp.ff_proj.weight": "model-00004-of-00004.safetensors",
|
| 220 |
+
"model.transformer.blocks.32.self_attn.att_proj.weight": "model-00003-of-00004.safetensors",
|
| 221 |
+
"model.transformer.blocks.32.self_attn.attn_out.weight": "model-00003-of-00004.safetensors",
|
| 222 |
+
"model.transformer.blocks.32.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
|
| 223 |
+
"model.transformer.blocks.32.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
|
| 224 |
+
"model.transformer.blocks.33.attn_norm.weight": "model-00004-of-00004.safetensors",
|
| 225 |
+
"model.transformer.blocks.33.ff_norm.weight": "model-00004-of-00004.safetensors",
|
| 226 |
+
"model.transformer.blocks.33.mlp.ff_out.weight": "model-00004-of-00004.safetensors",
|
| 227 |
+
"model.transformer.blocks.33.mlp.ff_proj.weight": "model-00004-of-00004.safetensors",
|
| 228 |
+
"model.transformer.blocks.33.self_attn.att_proj.weight": "model-00004-of-00004.safetensors",
|
| 229 |
+
"model.transformer.blocks.33.self_attn.attn_out.weight": "model-00004-of-00004.safetensors",
|
| 230 |
+
"model.transformer.blocks.33.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 231 |
+
"model.transformer.blocks.33.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 232 |
+
"model.transformer.blocks.34.attn_norm.weight": "model-00004-of-00004.safetensors",
|
| 233 |
+
"model.transformer.blocks.34.ff_norm.weight": "model-00004-of-00004.safetensors",
|
| 234 |
+
"model.transformer.blocks.34.mlp.ff_out.weight": "model-00004-of-00004.safetensors",
|
| 235 |
+
"model.transformer.blocks.34.mlp.ff_proj.weight": "model-00004-of-00004.safetensors",
|
| 236 |
+
"model.transformer.blocks.34.self_attn.att_proj.weight": "model-00004-of-00004.safetensors",
|
| 237 |
+
"model.transformer.blocks.34.self_attn.attn_out.weight": "model-00004-of-00004.safetensors",
|
| 238 |
+
"model.transformer.blocks.34.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 239 |
+
"model.transformer.blocks.34.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 240 |
+
"model.transformer.blocks.35.attn_norm.weight": "model-00004-of-00004.safetensors",
|
| 241 |
+
"model.transformer.blocks.35.ff_norm.weight": "model-00004-of-00004.safetensors",
|
| 242 |
+
"model.transformer.blocks.35.mlp.ff_out.weight": "model-00004-of-00004.safetensors",
|
| 243 |
+
"model.transformer.blocks.35.mlp.ff_proj.weight": "model-00004-of-00004.safetensors",
|
| 244 |
+
"model.transformer.blocks.35.self_attn.att_proj.weight": "model-00004-of-00004.safetensors",
|
| 245 |
+
"model.transformer.blocks.35.self_attn.attn_out.weight": "model-00004-of-00004.safetensors",
|
| 246 |
+
"model.transformer.blocks.35.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
|
| 247 |
+
"model.transformer.blocks.35.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
|
| 248 |
+
"model.transformer.blocks.4.attn_norm.weight": "model-00001-of-00004.safetensors",
|
| 249 |
+
"model.transformer.blocks.4.ff_norm.weight": "model-00001-of-00004.safetensors",
|
| 250 |
+
"model.transformer.blocks.4.mlp.ff_out.weight": "model-00001-of-00004.safetensors",
|
| 251 |
+
"model.transformer.blocks.4.mlp.ff_proj.weight": "model-00001-of-00004.safetensors",
|
| 252 |
+
"model.transformer.blocks.4.self_attn.att_proj.weight": "model-00001-of-00004.safetensors",
|
| 253 |
+
"model.transformer.blocks.4.self_attn.attn_out.weight": "model-00001-of-00004.safetensors",
|
| 254 |
+
"model.transformer.blocks.4.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 255 |
+
"model.transformer.blocks.4.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 256 |
+
"model.transformer.blocks.5.attn_norm.weight": "model-00001-of-00004.safetensors",
|
| 257 |
+
"model.transformer.blocks.5.ff_norm.weight": "model-00001-of-00004.safetensors",
|
| 258 |
+
"model.transformer.blocks.5.mlp.ff_out.weight": "model-00001-of-00004.safetensors",
|
| 259 |
+
"model.transformer.blocks.5.mlp.ff_proj.weight": "model-00001-of-00004.safetensors",
|
| 260 |
+
"model.transformer.blocks.5.self_attn.att_proj.weight": "model-00001-of-00004.safetensors",
|
| 261 |
+
"model.transformer.blocks.5.self_attn.attn_out.weight": "model-00001-of-00004.safetensors",
|
| 262 |
+
"model.transformer.blocks.5.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 263 |
+
"model.transformer.blocks.5.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 264 |
+
"model.transformer.blocks.6.attn_norm.weight": "model-00001-of-00004.safetensors",
|
| 265 |
+
"model.transformer.blocks.6.ff_norm.weight": "model-00001-of-00004.safetensors",
|
| 266 |
+
"model.transformer.blocks.6.mlp.ff_out.weight": "model-00001-of-00004.safetensors",
|
| 267 |
+
"model.transformer.blocks.6.mlp.ff_proj.weight": "model-00001-of-00004.safetensors",
|
| 268 |
+
"model.transformer.blocks.6.self_attn.att_proj.weight": "model-00001-of-00004.safetensors",
|
| 269 |
+
"model.transformer.blocks.6.self_attn.attn_out.weight": "model-00001-of-00004.safetensors",
|
| 270 |
+
"model.transformer.blocks.6.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 271 |
+
"model.transformer.blocks.6.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 272 |
+
"model.transformer.blocks.7.attn_norm.weight": "model-00001-of-00004.safetensors",
|
| 273 |
+
"model.transformer.blocks.7.ff_norm.weight": "model-00001-of-00004.safetensors",
|
| 274 |
+
"model.transformer.blocks.7.mlp.ff_out.weight": "model-00001-of-00004.safetensors",
|
| 275 |
+
"model.transformer.blocks.7.mlp.ff_proj.weight": "model-00001-of-00004.safetensors",
|
| 276 |
+
"model.transformer.blocks.7.self_attn.att_proj.weight": "model-00001-of-00004.safetensors",
|
| 277 |
+
"model.transformer.blocks.7.self_attn.attn_out.weight": "model-00001-of-00004.safetensors",
|
| 278 |
+
"model.transformer.blocks.7.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 279 |
+
"model.transformer.blocks.7.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 280 |
+
"model.transformer.blocks.8.attn_norm.weight": "model-00001-of-00004.safetensors",
|
| 281 |
+
"model.transformer.blocks.8.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 282 |
+
"model.transformer.blocks.8.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 283 |
+
"model.transformer.blocks.8.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 284 |
+
"model.transformer.blocks.8.self_attn.att_proj.weight": "model-00001-of-00004.safetensors",
|
| 285 |
+
"model.transformer.blocks.8.self_attn.attn_out.weight": "model-00001-of-00004.safetensors",
|
| 286 |
+
"model.transformer.blocks.8.self_attn.k_norm.weight": "model-00001-of-00004.safetensors",
|
| 287 |
+
"model.transformer.blocks.8.self_attn.q_norm.weight": "model-00001-of-00004.safetensors",
|
| 288 |
+
"model.transformer.blocks.9.attn_norm.weight": "model-00002-of-00004.safetensors",
|
| 289 |
+
"model.transformer.blocks.9.ff_norm.weight": "model-00002-of-00004.safetensors",
|
| 290 |
+
"model.transformer.blocks.9.mlp.ff_out.weight": "model-00002-of-00004.safetensors",
|
| 291 |
+
"model.transformer.blocks.9.mlp.ff_proj.weight": "model-00002-of-00004.safetensors",
|
| 292 |
+
"model.transformer.blocks.9.self_attn.att_proj.weight": "model-00002-of-00004.safetensors",
|
| 293 |
+
"model.transformer.blocks.9.self_attn.attn_out.weight": "model-00002-of-00004.safetensors",
|
| 294 |
+
"model.transformer.blocks.9.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
|
| 295 |
+
"model.transformer.blocks.9.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
|
| 296 |
+
"model.transformer.ln_f.weight": "model-00004-of-00004.safetensors",
|
| 297 |
+
"model.transformer.wte.embedding": "model-00001-of-00004.safetensors",
|
| 298 |
+
"model.transformer.wte.new_embedding": "model-00001-of-00004.safetensors",
|
| 299 |
+
"model.vision_backbone.image_pooling_2d.wk.bias": "model-00004-of-00004.safetensors",
|
| 300 |
+
"model.vision_backbone.image_pooling_2d.wk.weight": "model-00004-of-00004.safetensors",
|
| 301 |
+
"model.vision_backbone.image_pooling_2d.wo.bias": "model-00004-of-00004.safetensors",
|
| 302 |
+
"model.vision_backbone.image_pooling_2d.wo.weight": "model-00004-of-00004.safetensors",
|
| 303 |
+
"model.vision_backbone.image_pooling_2d.wq.bias": "model-00004-of-00004.safetensors",
|
| 304 |
+
"model.vision_backbone.image_pooling_2d.wq.weight": "model-00004-of-00004.safetensors",
|
| 305 |
+
"model.vision_backbone.image_pooling_2d.wv.bias": "model-00004-of-00004.safetensors",
|
| 306 |
+
"model.vision_backbone.image_pooling_2d.wv.weight": "model-00004-of-00004.safetensors",
|
| 307 |
+
"model.vision_backbone.image_projector.w1.weight": "model-00004-of-00004.safetensors",
|
| 308 |
+
"model.vision_backbone.image_projector.w2.weight": "model-00004-of-00004.safetensors",
|
| 309 |
+
"model.vision_backbone.image_projector.w3.weight": "model-00004-of-00004.safetensors",
|
| 310 |
+
"model.vision_backbone.image_vit.patch_embedding.bias": "model-00004-of-00004.safetensors",
|
| 311 |
+
"model.vision_backbone.image_vit.patch_embedding.weight": "model-00004-of-00004.safetensors",
|
| 312 |
+
"model.vision_backbone.image_vit.positional_embedding": "model-00004-of-00004.safetensors",
|
| 313 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 314 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 315 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 316 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 317 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 318 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 319 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 320 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 321 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 322 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 323 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 324 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 325 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 326 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 327 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 328 |
+
"model.vision_backbone.image_vit.transformer.resblocks.0.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 329 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 330 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 331 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 332 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 333 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 334 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 335 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 336 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 337 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 338 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 339 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 340 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 341 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 342 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 343 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 344 |
+
"model.vision_backbone.image_vit.transformer.resblocks.1.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 345 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 346 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 347 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 348 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 349 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 350 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 351 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 352 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 353 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 354 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 355 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 356 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 357 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 358 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 359 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 360 |
+
"model.vision_backbone.image_vit.transformer.resblocks.10.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 361 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 362 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 363 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 364 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 365 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 366 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 367 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 368 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 369 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 370 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 371 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 372 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 373 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 374 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 375 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 376 |
+
"model.vision_backbone.image_vit.transformer.resblocks.11.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 377 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 378 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 379 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 380 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 381 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 382 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 383 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 384 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 385 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 386 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 387 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 388 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 389 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 390 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 391 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 392 |
+
"model.vision_backbone.image_vit.transformer.resblocks.12.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 393 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 394 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 395 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 396 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 397 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 398 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 399 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 400 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 401 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 402 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 403 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 404 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 405 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 406 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 407 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 408 |
+
"model.vision_backbone.image_vit.transformer.resblocks.13.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 409 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 410 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 411 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 412 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 413 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 414 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 415 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 416 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 417 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 418 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 419 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 420 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 421 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 422 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 423 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 424 |
+
"model.vision_backbone.image_vit.transformer.resblocks.14.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 425 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 426 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 427 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 428 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 429 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 430 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 431 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 432 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 433 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 434 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 435 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 436 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 437 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 438 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 439 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 440 |
+
"model.vision_backbone.image_vit.transformer.resblocks.15.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 441 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 442 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 443 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 444 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 445 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 446 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 447 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 448 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 449 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 450 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 451 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 452 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 453 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 454 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 455 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 456 |
+
"model.vision_backbone.image_vit.transformer.resblocks.16.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 457 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 458 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 459 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 460 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 461 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 462 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 463 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 464 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 465 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 466 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 467 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 468 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 469 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 470 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 471 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 472 |
+
"model.vision_backbone.image_vit.transformer.resblocks.17.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 473 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 474 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 475 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 476 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 477 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 478 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 479 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 480 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 481 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 482 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 483 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 484 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 485 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 486 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 487 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 488 |
+
"model.vision_backbone.image_vit.transformer.resblocks.18.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 489 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 490 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 491 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 492 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 493 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 494 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 495 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 496 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 497 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 498 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 499 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 500 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 501 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 502 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 503 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 504 |
+
"model.vision_backbone.image_vit.transformer.resblocks.19.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 505 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 506 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 507 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 508 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 509 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 510 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 511 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 512 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 513 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 514 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 515 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 516 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 517 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 518 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 519 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 520 |
+
"model.vision_backbone.image_vit.transformer.resblocks.2.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 521 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 522 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 523 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 524 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 525 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 526 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 527 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 528 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 529 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 530 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 531 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 532 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 533 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 534 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 535 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 536 |
+
"model.vision_backbone.image_vit.transformer.resblocks.20.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 537 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 538 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 539 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 540 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 541 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 542 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 543 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 544 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 545 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 546 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 547 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 548 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 549 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 550 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 551 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 552 |
+
"model.vision_backbone.image_vit.transformer.resblocks.21.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 553 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 554 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 555 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 556 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 557 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 558 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 559 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 560 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 561 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 562 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 563 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 564 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 565 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 566 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 567 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 568 |
+
"model.vision_backbone.image_vit.transformer.resblocks.22.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 569 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 570 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 571 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 572 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 573 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 574 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 575 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 576 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 577 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 578 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 579 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 580 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 581 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 582 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 583 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 584 |
+
"model.vision_backbone.image_vit.transformer.resblocks.23.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 585 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 586 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 587 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 588 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 589 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 590 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 591 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 592 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 593 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 594 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 595 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 596 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 597 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 598 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 599 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 600 |
+
"model.vision_backbone.image_vit.transformer.resblocks.24.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 601 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 602 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 603 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 604 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 605 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 606 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 607 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 608 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 609 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 610 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 611 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 612 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 613 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 614 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 615 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 616 |
+
"model.vision_backbone.image_vit.transformer.resblocks.3.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 617 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 618 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 619 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 620 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 621 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 622 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 623 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 624 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 625 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 626 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 627 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 628 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 629 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 630 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 631 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 632 |
+
"model.vision_backbone.image_vit.transformer.resblocks.4.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 633 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 634 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 635 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 636 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 637 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 638 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 639 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 640 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 641 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 642 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 643 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 644 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 645 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 646 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 647 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 648 |
+
"model.vision_backbone.image_vit.transformer.resblocks.5.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 649 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 650 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 651 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 652 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 653 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 654 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 655 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 656 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 657 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 658 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 659 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 660 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 661 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 662 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 663 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 664 |
+
"model.vision_backbone.image_vit.transformer.resblocks.6.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 665 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 666 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 667 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 668 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 669 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 670 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 671 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 672 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 673 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 674 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 675 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 676 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 677 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 678 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 679 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 680 |
+
"model.vision_backbone.image_vit.transformer.resblocks.7.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 681 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 682 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 683 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 684 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 685 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 686 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 687 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 688 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 689 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 690 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 691 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 692 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 693 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 694 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 695 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 696 |
+
"model.vision_backbone.image_vit.transformer.resblocks.8.ffn_norm.weight": "model-00004-of-00004.safetensors",
|
| 697 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.attention.wk.bias": "model-00004-of-00004.safetensors",
|
| 698 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.attention.wk.weight": "model-00004-of-00004.safetensors",
|
| 699 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.attention.wo.bias": "model-00004-of-00004.safetensors",
|
| 700 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.attention.wo.weight": "model-00004-of-00004.safetensors",
|
| 701 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.attention.wq.bias": "model-00004-of-00004.safetensors",
|
| 702 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.attention.wq.weight": "model-00004-of-00004.safetensors",
|
| 703 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.attention.wv.bias": "model-00004-of-00004.safetensors",
|
| 704 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.attention.wv.weight": "model-00004-of-00004.safetensors",
|
| 705 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.attention_norm.bias": "model-00004-of-00004.safetensors",
|
| 706 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.attention_norm.weight": "model-00004-of-00004.safetensors",
|
| 707 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.feed_forward.w1.bias": "model-00004-of-00004.safetensors",
|
| 708 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
|
| 709 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.feed_forward.w2.bias": "model-00004-of-00004.safetensors",
|
| 710 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
|
| 711 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.ffn_norm.bias": "model-00004-of-00004.safetensors",
|
| 712 |
+
"model.vision_backbone.image_vit.transformer.resblocks.9.ffn_norm.weight": "model-00004-of-00004.safetensors"
|
| 713 |
+
}
|
| 714 |
+
}
|
modeling_molmo2.py
ADDED
|
@@ -0,0 +1,1764 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import math
|
| 2 |
+
from copy import deepcopy
|
| 3 |
+
from dataclasses import dataclass
|
| 4 |
+
from typing import Optional, Union, Callable
|
| 5 |
+
|
| 6 |
+
import torch
|
| 7 |
+
from torch import nn
|
| 8 |
+
from torch.nn import functional as F
|
| 9 |
+
|
| 10 |
+
from transformers.models.auto import AutoModelForImageTextToText
|
| 11 |
+
from transformers.activations import ACT2FN
|
| 12 |
+
from transformers.configuration_utils import PretrainedConfig
|
| 13 |
+
from transformers.cache_utils import Cache, DynamicCache
|
| 14 |
+
from transformers.generation import GenerationMixin
|
| 15 |
+
from transformers.masking_utils import create_causal_mask, create_masks_for_generate
|
| 16 |
+
from transformers.modeling_flash_attention_utils import (
|
| 17 |
+
_flash_attention_forward,
|
| 18 |
+
FlashAttentionKwargs,
|
| 19 |
+
flash_attn_supports_top_left_mask,
|
| 20 |
+
)
|
| 21 |
+
from transformers.modeling_layers import GradientCheckpointingLayer
|
| 22 |
+
from transformers.modeling_outputs import (
|
| 23 |
+
BaseModelOutputWithPast,
|
| 24 |
+
)
|
| 25 |
+
from transformers.modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update
|
| 26 |
+
from transformers.modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel
|
| 27 |
+
from transformers.processing_utils import Unpack
|
| 28 |
+
from transformers.utils import (
|
| 29 |
+
ModelOutput,
|
| 30 |
+
TransformersKwargs,
|
| 31 |
+
can_return_tuple,
|
| 32 |
+
logging,
|
| 33 |
+
)
|
| 34 |
+
|
| 35 |
+
from .configuration_molmo2 import Molmo2Config, Molmo2VitConfig, Molmo2AdapterConfig, Molmo2TextConfig
|
| 36 |
+
|
| 37 |
+
|
| 38 |
+
logger = logging.get_logger(__name__)
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
@dataclass
|
| 42 |
+
class Molmo2CausalLMOutputWithPast(ModelOutput):
|
| 43 |
+
"""
|
| 44 |
+
Base class for Molmo2 causal language model (or autoregressive) outputs.
|
| 45 |
+
|
| 46 |
+
Args:
|
| 47 |
+
loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided):
|
| 48 |
+
Language modeling loss (for next-token prediction).
|
| 49 |
+
logits (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`):
|
| 50 |
+
Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
|
| 51 |
+
past_key_values (`Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
|
| 52 |
+
It is a [`~cache_utils.Cache`] instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache).
|
| 53 |
+
|
| 54 |
+
Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see
|
| 55 |
+
`past_key_values` input) to speed up sequential decoding.
|
| 56 |
+
image_hidden_states (`torch.FloatTensor`, *optional*):
|
| 57 |
+
A `torch.FloatTensor` of size `(batch_size, num_images, sequence_length, hidden_size)`.
|
| 58 |
+
image_hidden_states of the model produced by the vision encoder and after projecting the last hidden state.
|
| 59 |
+
"""
|
| 60 |
+
|
| 61 |
+
loss: Optional[torch.FloatTensor] = None
|
| 62 |
+
logits: Optional[torch.FloatTensor] = None
|
| 63 |
+
past_key_values: Optional[Cache] = None
|
| 64 |
+
hidden_states: Optional[tuple[torch.FloatTensor]] = None
|
| 65 |
+
attentions: Optional[tuple[torch.FloatTensor]] = None
|
| 66 |
+
image_hidden_states: Optional[torch.FloatTensor] = None
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
@dataclass
|
| 70 |
+
class Molmo2ModelOutputWithPast(BaseModelOutputWithPast):
|
| 71 |
+
"""
|
| 72 |
+
Base class for Molmo2 outputs, with hidden states and attentions.
|
| 73 |
+
|
| 74 |
+
Args:
|
| 75 |
+
image_hidden_states (`torch.FloatTensor`, *optional*):
|
| 76 |
+
A `torch.FloatTensor` of size `(batch_num_patches, hidden_size)`.
|
| 77 |
+
image_hidden_states of the model produced by the vision backbone
|
| 78 |
+
"""
|
| 79 |
+
last_hidden_state: Optional[torch.FloatTensor] = None
|
| 80 |
+
past_key_values: Optional[Cache] = None
|
| 81 |
+
hidden_states: Optional[tuple[torch.FloatTensor]] = None
|
| 82 |
+
attentions: Optional[tuple[torch.FloatTensor]] = None
|
| 83 |
+
image_hidden_states: Optional[torch.FloatTensor] = None
|
| 84 |
+
|
| 85 |
+
|
| 86 |
+
class ViTMLP(nn.Module):
|
| 87 |
+
def __init__(self, dim: int, hidden_dim: int, hidden_act: str, device: Union[str, torch.device] = None):
|
| 88 |
+
super().__init__()
|
| 89 |
+
self.w1 = nn.Linear(dim, hidden_dim, bias=True, device=device)
|
| 90 |
+
self.act = ACT2FN[hidden_act]
|
| 91 |
+
self.w2 = nn.Linear(hidden_dim, dim, bias=True, device=device)
|
| 92 |
+
|
| 93 |
+
def forward(self, x: torch.Tensor) -> torch.Tensor:
|
| 94 |
+
return self.w2(self.act(self.w1(x)))
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
class ViTMultiHeadDotProductAttention(nn.Module):
|
| 98 |
+
def __init__(
|
| 99 |
+
self,
|
| 100 |
+
hidden_size: int,
|
| 101 |
+
num_heads: int,
|
| 102 |
+
num_key_value_heads: int,
|
| 103 |
+
head_dim: int,
|
| 104 |
+
use_bias: bool = True,
|
| 105 |
+
input_dim: Optional[int] = None,
|
| 106 |
+
float32_attention: bool = True,
|
| 107 |
+
attention_dropout: float = 0.0,
|
| 108 |
+
residual_dropout: float = 0.0,
|
| 109 |
+
device: Union[str, torch.device] = None,
|
| 110 |
+
attn_implementation: str = "eager",
|
| 111 |
+
):
|
| 112 |
+
super().__init__()
|
| 113 |
+
|
| 114 |
+
self.hidden_size = hidden_size
|
| 115 |
+
self.num_heads = num_heads
|
| 116 |
+
self.head_dim = head_dim
|
| 117 |
+
self.num_key_value_heads = num_key_value_heads
|
| 118 |
+
self.num_key_value_groups = self.num_heads // self.num_key_value_heads
|
| 119 |
+
self.attn_implementation = attn_implementation
|
| 120 |
+
self.is_causal = False
|
| 121 |
+
|
| 122 |
+
input_dim = input_dim or hidden_size
|
| 123 |
+
|
| 124 |
+
self.wq = nn.Linear(
|
| 125 |
+
input_dim,
|
| 126 |
+
self.num_heads * self.head_dim,
|
| 127 |
+
bias=use_bias,
|
| 128 |
+
device=device,
|
| 129 |
+
)
|
| 130 |
+
self.wk = nn.Linear(
|
| 131 |
+
input_dim,
|
| 132 |
+
self.num_key_value_heads * self.head_dim,
|
| 133 |
+
bias=use_bias,
|
| 134 |
+
device=device,
|
| 135 |
+
)
|
| 136 |
+
self.wv = nn.Linear(
|
| 137 |
+
input_dim,
|
| 138 |
+
self.num_key_value_heads * self.head_dim,
|
| 139 |
+
bias=use_bias,
|
| 140 |
+
device=device,
|
| 141 |
+
)
|
| 142 |
+
self.wo = nn.Linear(
|
| 143 |
+
self.num_heads * self.head_dim,
|
| 144 |
+
self.hidden_size,
|
| 145 |
+
)
|
| 146 |
+
self.float32_attention = float32_attention
|
| 147 |
+
self.attention_dropout = attention_dropout
|
| 148 |
+
self.residual_dropout = nn.Dropout(residual_dropout)
|
| 149 |
+
|
| 150 |
+
def _split_heads(self, hidden_states, num_heads) -> torch.Tensor:
|
| 151 |
+
return hidden_states.reshape(hidden_states.shape[:2] + (num_heads, self.head_dim))
|
| 152 |
+
|
| 153 |
+
def _merge_heads(self, hidden_states) -> torch.Tensor:
|
| 154 |
+
return hidden_states.reshape(hidden_states.shape[:2] + (self.hidden_size,))
|
| 155 |
+
|
| 156 |
+
def forward(
|
| 157 |
+
self,
|
| 158 |
+
inputs_q: torch.Tensor,
|
| 159 |
+
inputs_kv: Optional[torch.Tensor] = None,
|
| 160 |
+
attn_mask: Optional[torch.Tensor] = None,
|
| 161 |
+
) -> torch.Tensor:
|
| 162 |
+
|
| 163 |
+
if inputs_kv is not None:
|
| 164 |
+
inputs_k = inputs_kv
|
| 165 |
+
inputs_v = inputs_kv
|
| 166 |
+
else:
|
| 167 |
+
inputs_k = inputs_q
|
| 168 |
+
inputs_v = inputs_q
|
| 169 |
+
|
| 170 |
+
xq, xk, xv = self.wq(inputs_q), self.wk(inputs_k), self.wv(inputs_v)
|
| 171 |
+
|
| 172 |
+
xq = self._split_heads(xq, self.num_heads)
|
| 173 |
+
xk = self._split_heads(xk, self.num_key_value_heads)
|
| 174 |
+
xv = self._split_heads(xv, self.num_key_value_heads)
|
| 175 |
+
|
| 176 |
+
if self.num_heads != self.num_key_value_heads:
|
| 177 |
+
xk = xk.repeat_interleave(self.num_key_value_groups, dim=2, output_size=self.num_heads)
|
| 178 |
+
xv = xv.repeat_interleave(self.num_key_value_groups, dim=2, output_size=self.num_heads)
|
| 179 |
+
|
| 180 |
+
og_dtype = xq.dtype
|
| 181 |
+
|
| 182 |
+
if self.float32_attention:
|
| 183 |
+
xq = xq.to(torch.float)
|
| 184 |
+
xk = xk.to(torch.float)
|
| 185 |
+
|
| 186 |
+
dropout_p = 0.0 if not self.training else self.attention_dropout
|
| 187 |
+
|
| 188 |
+
if self.attn_implementation == "eager":
|
| 189 |
+
attn_weights = torch.einsum("...qhd,...khd->...hqk", xq / math.sqrt(xq.size(-1)), xk)
|
| 190 |
+
attn_weights = F.softmax(attn_weights, dim=-1, dtype=torch.float32).to(xq.dtype)
|
| 191 |
+
attn_weights = F.dropout(
|
| 192 |
+
attn_weights,
|
| 193 |
+
p=dropout_p,
|
| 194 |
+
training=self.training
|
| 195 |
+
)
|
| 196 |
+
attn_output = torch.einsum("...hqk,...khd->...qhd", attn_weights.to(xv.dtype), xv)
|
| 197 |
+
|
| 198 |
+
elif self.attn_implementation == "sdpa":
|
| 199 |
+
if not torch.is_autocast_enabled():
|
| 200 |
+
xv = xv.to(torch.float)
|
| 201 |
+
|
| 202 |
+
attn_output = F.scaled_dot_product_attention(
|
| 203 |
+
xq.transpose(1, 2).contiguous(),
|
| 204 |
+
xk.transpose(1, 2).contiguous(),
|
| 205 |
+
xv.transpose(1, 2).contiguous(),
|
| 206 |
+
attn_mask=attn_mask,
|
| 207 |
+
is_causal=False,
|
| 208 |
+
dropout_p=dropout_p,
|
| 209 |
+
).transpose(1, 2)
|
| 210 |
+
|
| 211 |
+
elif self.attn_implementation == "flash_attention_2":
|
| 212 |
+
if xq.dtype == torch.float32:
|
| 213 |
+
if torch.is_autocast_enabled():
|
| 214 |
+
target_dtype = torch.get_autocast_gpu_dtype()
|
| 215 |
+
else:
|
| 216 |
+
target_dtype = self.wq.weight.dtype
|
| 217 |
+
attn_output = _flash_attention_forward(
|
| 218 |
+
xq,
|
| 219 |
+
xk,
|
| 220 |
+
xv,
|
| 221 |
+
attention_mask=attn_mask,
|
| 222 |
+
query_length=inputs_q.shape[1],
|
| 223 |
+
is_causal=False,
|
| 224 |
+
dropout=dropout_p,
|
| 225 |
+
softmax_scale=xq.shape[-1] ** -0.5,
|
| 226 |
+
use_top_left_mask=flash_attn_supports_top_left_mask(),
|
| 227 |
+
target_dtype=target_dtype,
|
| 228 |
+
implementation=self.attn_implementation,
|
| 229 |
+
)
|
| 230 |
+
else:
|
| 231 |
+
raise ValueError(f"Attention implementation {self.attn_implementation} not supported")
|
| 232 |
+
|
| 233 |
+
attn_output = attn_output.to(og_dtype)
|
| 234 |
+
attn_output = self._merge_heads(attn_output)
|
| 235 |
+
attn_output = self.wo(attn_output)
|
| 236 |
+
attn_output = self.residual_dropout(attn_output)
|
| 237 |
+
|
| 238 |
+
return attn_output
|
| 239 |
+
|
| 240 |
+
|
| 241 |
+
class Molmo2VisionBlock(nn.Module):
|
| 242 |
+
|
| 243 |
+
def __init__(self, config: Molmo2VitConfig, device: Union[str, torch.device] = None):
|
| 244 |
+
super().__init__()
|
| 245 |
+
self.attention = ViTMultiHeadDotProductAttention(
|
| 246 |
+
hidden_size=config.hidden_size,
|
| 247 |
+
num_heads=config.num_attention_heads,
|
| 248 |
+
num_key_value_heads=config.num_key_value_heads,
|
| 249 |
+
head_dim=config.head_dim,
|
| 250 |
+
float32_attention=config.float32_attention,
|
| 251 |
+
attention_dropout=config.attention_dropout,
|
| 252 |
+
residual_dropout=config.residual_dropout,
|
| 253 |
+
device=device,
|
| 254 |
+
attn_implementation=config._attn_implementation,
|
| 255 |
+
)
|
| 256 |
+
self.feed_forward = ViTMLP(config.hidden_size, config.intermediate_size, config.hidden_act, device=device)
|
| 257 |
+
self.attention_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps, device=device)
|
| 258 |
+
self.ffn_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps, device=device)
|
| 259 |
+
|
| 260 |
+
def forward(self, x: torch.Tensor) -> torch.Tensor:
|
| 261 |
+
x = x + self.attention(self.attention_norm(x))
|
| 262 |
+
x = x + self.feed_forward(self.ffn_norm(x))
|
| 263 |
+
return x
|
| 264 |
+
|
| 265 |
+
|
| 266 |
+
class Molmo2VisionBlockCollection(nn.Module):
|
| 267 |
+
|
| 268 |
+
def __init__(self, config: Molmo2VitConfig, device: Union[str, torch.device] = None):
|
| 269 |
+
super().__init__()
|
| 270 |
+
self.conifg = config
|
| 271 |
+
self.resblocks = nn.ModuleList([
|
| 272 |
+
Molmo2VisionBlock(config, device) for _ in range(config.num_hidden_layers)
|
| 273 |
+
])
|
| 274 |
+
|
| 275 |
+
def forward(self, x: torch.Tensor) -> list[torch.Tensor]:
|
| 276 |
+
hidden_states = []
|
| 277 |
+
for r in self.resblocks:
|
| 278 |
+
x = r(x)
|
| 279 |
+
hidden_states.append(x)
|
| 280 |
+
return hidden_states
|
| 281 |
+
|
| 282 |
+
|
| 283 |
+
class Molmo2VisionTransformer(nn.Module):
|
| 284 |
+
|
| 285 |
+
def __init__(self, config: Molmo2VitConfig, device: Union[str, torch.device] = None):
|
| 286 |
+
super().__init__()
|
| 287 |
+
self.config = config
|
| 288 |
+
|
| 289 |
+
# positional embeddings
|
| 290 |
+
self.scale = config.hidden_size ** -0.5
|
| 291 |
+
self.num_prefix_tokens: int = 0 # no class embeddings
|
| 292 |
+
self.positional_embedding = nn.Parameter(
|
| 293 |
+
torch.zeros(config.image_num_pos, config.hidden_size, device=device),
|
| 294 |
+
)
|
| 295 |
+
|
| 296 |
+
image_patch_size = config.image_patch_size
|
| 297 |
+
self.patch_embedding = nn.Linear(
|
| 298 |
+
image_patch_size * image_patch_size * 3,
|
| 299 |
+
config.hidden_size,
|
| 300 |
+
bias=True,
|
| 301 |
+
device=device,
|
| 302 |
+
)
|
| 303 |
+
|
| 304 |
+
self.transformer = Molmo2VisionBlockCollection(config, device)
|
| 305 |
+
|
| 306 |
+
def add_pos_emb(self, x: torch.Tensor, patch_num: int) -> torch.Tensor:
|
| 307 |
+
pos_emb = self.positional_embedding
|
| 308 |
+
|
| 309 |
+
pos_emb = pos_emb.reshape(
|
| 310 |
+
(int(math.sqrt(pos_emb.shape[0])), int(math.sqrt(pos_emb.shape[0])), pos_emb.shape[1])
|
| 311 |
+
)
|
| 312 |
+
|
| 313 |
+
(patch_num_0, patch_num_1) = patch_num
|
| 314 |
+
|
| 315 |
+
if pos_emb.shape[0] != patch_num_0 or pos_emb.shape[1] != patch_num_1:
|
| 316 |
+
# Dervied from https://github.com/facebookresearch/mae/blob/main/util/pos_embed.py
|
| 317 |
+
# antialias: default True in jax.image.resize
|
| 318 |
+
pos_emb = pos_emb.unsqueeze(0).permute(0, 3, 1, 2)
|
| 319 |
+
pos_emb = F.interpolate(
|
| 320 |
+
pos_emb, size=(patch_num_0, patch_num_1), mode="bicubic", align_corners=False, antialias=True,
|
| 321 |
+
)
|
| 322 |
+
pos_emb = pos_emb.permute(0, 2, 3, 1).squeeze(0)
|
| 323 |
+
|
| 324 |
+
pos_emb = pos_emb.reshape(-1, pos_emb.shape[-1])
|
| 325 |
+
x = x + pos_emb[None, :, :].to(x.dtype)
|
| 326 |
+
return x
|
| 327 |
+
|
| 328 |
+
def forward(self, x: torch.Tensor, patch_num: int = None) -> list[torch.Tensor]:
|
| 329 |
+
"""
|
| 330 |
+
: param x: (batch_size, num_patch, n_pixels)
|
| 331 |
+
"""
|
| 332 |
+
if patch_num is None:
|
| 333 |
+
patch_num = self.config.image_num_patch
|
| 334 |
+
|
| 335 |
+
B, N, D = x.shape
|
| 336 |
+
|
| 337 |
+
x = self.patch_embedding(x)
|
| 338 |
+
|
| 339 |
+
# class embeddings and positional embeddings
|
| 340 |
+
x = self.add_pos_emb(x, patch_num)
|
| 341 |
+
|
| 342 |
+
hidden_states = self.transformer(x)
|
| 343 |
+
return hidden_states
|
| 344 |
+
|
| 345 |
+
|
| 346 |
+
class ImageProjectorMLP(nn.Module):
|
| 347 |
+
|
| 348 |
+
def __init__(
|
| 349 |
+
self,
|
| 350 |
+
input_dim: int,
|
| 351 |
+
hidden_dim: int,
|
| 352 |
+
output_dim: int,
|
| 353 |
+
hidden_act: str,
|
| 354 |
+
device: Union[str, torch.device] = None,
|
| 355 |
+
):
|
| 356 |
+
super().__init__()
|
| 357 |
+
self.w1 = nn.Linear(input_dim, hidden_dim, bias=False, device=device)
|
| 358 |
+
self.w2 = nn.Linear(hidden_dim, output_dim, bias=False, device=device)
|
| 359 |
+
self.w3 = nn.Linear(input_dim, hidden_dim, bias=False, device=device)
|
| 360 |
+
self.act = ACT2FN[hidden_act]
|
| 361 |
+
|
| 362 |
+
def forward(self, x: torch.Tensor) -> torch.Tensor:
|
| 363 |
+
return self.w2(self.act(self.w1(x)) * self.w3(x))
|
| 364 |
+
|
| 365 |
+
|
| 366 |
+
class Molmo2VisionBackbone(nn.Module):
|
| 367 |
+
def __init__(self, vit_config: Molmo2VitConfig, adapter_config: Molmo2AdapterConfig):
|
| 368 |
+
super().__init__()
|
| 369 |
+
self.vit_config = vit_config
|
| 370 |
+
self.adapter_config = adapter_config
|
| 371 |
+
|
| 372 |
+
self.vit_layers = []
|
| 373 |
+
for layer in adapter_config.vit_layers:
|
| 374 |
+
if layer >= 0:
|
| 375 |
+
self.vit_layers.append(layer)
|
| 376 |
+
else:
|
| 377 |
+
self.vit_layers.append(layer + vit_config.num_hidden_layers)
|
| 378 |
+
|
| 379 |
+
last_layer_needed = max(self.vit_layers) + 1
|
| 380 |
+
if last_layer_needed < vit_config.num_hidden_layers:
|
| 381 |
+
new_vit_config = deepcopy(vit_config)
|
| 382 |
+
new_vit_config.num_hidden_layers = last_layer_needed
|
| 383 |
+
self.image_vit = Molmo2VisionTransformer(new_vit_config)
|
| 384 |
+
else:
|
| 385 |
+
self.image_vit = Molmo2VisionTransformer(vit_config)
|
| 386 |
+
|
| 387 |
+
self.num_prefix_tokens: int = self.image_vit.num_prefix_tokens
|
| 388 |
+
|
| 389 |
+
pool_dim = vit_config.hidden_size * len(adapter_config.vit_layers)
|
| 390 |
+
self.image_pooling_2d = ViTMultiHeadDotProductAttention(
|
| 391 |
+
hidden_size=adapter_config.hidden_size,
|
| 392 |
+
num_heads=adapter_config.num_attention_heads,
|
| 393 |
+
num_key_value_heads=adapter_config.num_key_value_heads,
|
| 394 |
+
head_dim=adapter_config.head_dim,
|
| 395 |
+
input_dim=pool_dim,
|
| 396 |
+
float32_attention=adapter_config.float32_attention,
|
| 397 |
+
attention_dropout=adapter_config.attention_dropout,
|
| 398 |
+
residual_dropout=adapter_config.residual_dropout,
|
| 399 |
+
attn_implementation=adapter_config._attn_implementation,
|
| 400 |
+
)
|
| 401 |
+
self.image_projector = ImageProjectorMLP(
|
| 402 |
+
adapter_config.hidden_size,
|
| 403 |
+
adapter_config.intermediate_size,
|
| 404 |
+
adapter_config.text_hidden_size,
|
| 405 |
+
adapter_config.hidden_act,
|
| 406 |
+
)
|
| 407 |
+
self.image_feature_dropout = nn.Dropout(adapter_config.image_feature_dropout)
|
| 408 |
+
|
| 409 |
+
def encode_image(self, images: torch.Tensor) -> torch.Tensor:
|
| 410 |
+
"""
|
| 411 |
+
: param images: (batch_size, num_crops, num_patch, n_pixels)
|
| 412 |
+
"""
|
| 413 |
+
B, T, N, D = images.shape
|
| 414 |
+
images = images.view(B * T, N, D)
|
| 415 |
+
image_features = self.image_vit(images)
|
| 416 |
+
|
| 417 |
+
features = []
|
| 418 |
+
for layer in self.vit_layers:
|
| 419 |
+
features.append(image_features[layer])
|
| 420 |
+
image_features = torch.cat(features, dim=-1)
|
| 421 |
+
|
| 422 |
+
if self.num_prefix_tokens > 0:
|
| 423 |
+
image_features = image_features[:, 1:]
|
| 424 |
+
image_features = image_features.view(B, T, N, -1)
|
| 425 |
+
return image_features
|
| 426 |
+
|
| 427 |
+
@property
|
| 428 |
+
def dtype(self) -> torch.dtype:
|
| 429 |
+
return self.image_vit.patch_embedding.weight.dtype
|
| 430 |
+
|
| 431 |
+
@property
|
| 432 |
+
def device(self) -> torch.device:
|
| 433 |
+
return self.image_vit.patch_embedding.weight.device
|
| 434 |
+
|
| 435 |
+
def forward(
|
| 436 |
+
self,
|
| 437 |
+
images: torch.Tensor,
|
| 438 |
+
pooled_patches_idx: torch.Tensor,
|
| 439 |
+
) -> tuple[torch.Tensor, Optional[torch.Tensor]]:
|
| 440 |
+
|
| 441 |
+
# image_features: (batch_size, num_crops(=num_image), num_patch, nximage_emb_dim)
|
| 442 |
+
batch_size, num_image = images.shape[:2]
|
| 443 |
+
images = images.to(device=self.device, dtype=self.dtype)
|
| 444 |
+
image_features = self.encode_image(images)
|
| 445 |
+
|
| 446 |
+
image_features = self.image_feature_dropout(image_features)
|
| 447 |
+
dim = image_features.shape[-1]
|
| 448 |
+
valid = pooled_patches_idx >= 0
|
| 449 |
+
valid_token = torch.any(valid, -1)
|
| 450 |
+
|
| 451 |
+
# Use `pooled_patches_idx` to arange the features for image pooling
|
| 452 |
+
batch_idx = torch.arange(pooled_patches_idx.shape[0], dtype=torch.long, device=pooled_patches_idx.device)
|
| 453 |
+
batch_idx = torch.tile(batch_idx.view(batch_size, 1, 1), [1, pooled_patches_idx.shape[1], pooled_patches_idx.shape[2]])
|
| 454 |
+
|
| 455 |
+
# Now [batch, num_high_res_features, pool_dim, dim]
|
| 456 |
+
to_pool = image_features.reshape(batch_size, -1, dim)[batch_idx, torch.clip(pooled_patches_idx, 0)]
|
| 457 |
+
to_pool = to_pool * valid.to(self.dtype)[:, :, :, None]
|
| 458 |
+
to_pool = to_pool.reshape([-1, pooled_patches_idx.shape[-1], dim])
|
| 459 |
+
if self.adapter_config.pooling_attention_mask:
|
| 460 |
+
attn_mask = valid.reshape([-1, 1, 1, valid.shape[-1]])
|
| 461 |
+
denom = valid.view(-1, to_pool.shape[-2]).float().sum(-1)
|
| 462 |
+
denom = torch.where(denom == 0, 1, denom)
|
| 463 |
+
query = to_pool.sum(-2, keepdim=True) / denom[:, None, None].to(to_pool.dtype)
|
| 464 |
+
else:
|
| 465 |
+
attn_mask = None
|
| 466 |
+
query = to_pool.mean(-2, keepdim=True)
|
| 467 |
+
pooled_features = self.image_pooling_2d(query, to_pool, attn_mask=attn_mask)
|
| 468 |
+
pooled_features = pooled_features.reshape([batch_size, -1, pooled_features.shape[-1]])
|
| 469 |
+
|
| 470 |
+
# MLP layer to map the feature.
|
| 471 |
+
pooled_features = self.image_projector(pooled_features)
|
| 472 |
+
return pooled_features.view(-1, pooled_features.shape[-1])[valid_token.flatten()]
|
| 473 |
+
|
| 474 |
+
|
| 475 |
+
# Copied from transformers.models.llama.modeling_llama.rotate_half
|
| 476 |
+
def rotate_half(x):
|
| 477 |
+
"""Rotates half the hidden dims of the input."""
|
| 478 |
+
x1 = x[..., : x.shape[-1] // 2]
|
| 479 |
+
x2 = x[..., x.shape[-1] // 2 :]
|
| 480 |
+
return torch.cat((-x2, x1), dim=-1)
|
| 481 |
+
|
| 482 |
+
|
| 483 |
+
# Copied from transformers.models.llama.modeling_llama.apply_rotary_pos_emb
|
| 484 |
+
def apply_rotary_pos_emb(q, k, cos, sin, position_ids=None, unsqueeze_dim=1):
|
| 485 |
+
"""Applies Rotary Position Embedding to the query and key tensors.
|
| 486 |
+
|
| 487 |
+
Args:
|
| 488 |
+
q (`torch.Tensor`): The query tensor.
|
| 489 |
+
k (`torch.Tensor`): The key tensor.
|
| 490 |
+
cos (`torch.Tensor`): The cosine part of the rotary embedding.
|
| 491 |
+
sin (`torch.Tensor`): The sine part of the rotary embedding.
|
| 492 |
+
position_ids (`torch.Tensor`, *optional*):
|
| 493 |
+
Deprecated and unused.
|
| 494 |
+
unsqueeze_dim (`int`, *optional*, defaults to 1):
|
| 495 |
+
The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and
|
| 496 |
+
sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note
|
| 497 |
+
that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and
|
| 498 |
+
k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes
|
| 499 |
+
cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have
|
| 500 |
+
the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2.
|
| 501 |
+
Returns:
|
| 502 |
+
`tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding.
|
| 503 |
+
"""
|
| 504 |
+
cos = cos.unsqueeze(unsqueeze_dim)
|
| 505 |
+
sin = sin.unsqueeze(unsqueeze_dim)
|
| 506 |
+
q_embed = (q * cos) + (rotate_half(q) * sin)
|
| 507 |
+
k_embed = (k * cos) + (rotate_half(k) * sin)
|
| 508 |
+
return q_embed, k_embed
|
| 509 |
+
|
| 510 |
+
|
| 511 |
+
class Molmo2RotaryEmbedding(nn.Module):
|
| 512 |
+
inv_freq: torch.Tensor # fix linting for `register_buffer`
|
| 513 |
+
|
| 514 |
+
def __init__(
|
| 515 |
+
self,
|
| 516 |
+
config: Molmo2TextConfig,
|
| 517 |
+
device: Union[str, torch.device] = None,
|
| 518 |
+
rope_type: Optional[str] = None,
|
| 519 |
+
):
|
| 520 |
+
super().__init__()
|
| 521 |
+
if rope_type is not None:
|
| 522 |
+
self.rope_type = rope_type
|
| 523 |
+
elif hasattr(config, "rope_scaling") and isinstance(config.rope_scaling, dict):
|
| 524 |
+
# BC: "rope_type" was originally "type"
|
| 525 |
+
self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling.get("type"))
|
| 526 |
+
else:
|
| 527 |
+
self.rope_type = "default"
|
| 528 |
+
self.max_seq_len_cached = config.max_position_embeddings
|
| 529 |
+
self.original_max_seq_len = config.max_position_embeddings
|
| 530 |
+
|
| 531 |
+
self.config = config
|
| 532 |
+
self.rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type]
|
| 533 |
+
|
| 534 |
+
inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device)
|
| 535 |
+
self.register_buffer("inv_freq", inv_freq, persistent=False)
|
| 536 |
+
self.original_inv_freq = self.inv_freq
|
| 537 |
+
|
| 538 |
+
@torch.no_grad()
|
| 539 |
+
@dynamic_rope_update # power user: used with advanced RoPE types (e.g. dynamic rope)
|
| 540 |
+
def forward(self, x, position_ids: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]:
|
| 541 |
+
inv_freq_expanded = self.inv_freq[None, :, None].float().expand(position_ids.shape[0], -1, 1).to(x.device)
|
| 542 |
+
position_ids_expanded = position_ids[:, None, :].float()
|
| 543 |
+
|
| 544 |
+
device_type = x.device.type if isinstance(x.device.type, str) and x.device.type != "mps" else "cpu"
|
| 545 |
+
with torch.autocast(device_type=device_type, enabled=False): # Force float32
|
| 546 |
+
freqs = (inv_freq_expanded.float() @ position_ids_expanded.float()).transpose(1, 2)
|
| 547 |
+
emb = torch.cat((freqs, freqs), dim=-1)
|
| 548 |
+
cos = emb.cos() * self.attention_scaling
|
| 549 |
+
sin = emb.sin() * self.attention_scaling
|
| 550 |
+
|
| 551 |
+
return cos.to(dtype=x.dtype), sin.to(dtype=x.dtype)
|
| 552 |
+
|
| 553 |
+
|
| 554 |
+
class Molmo2RMSNorm(nn.Module):
|
| 555 |
+
|
| 556 |
+
def __init__(
|
| 557 |
+
self,
|
| 558 |
+
size: int,
|
| 559 |
+
eps: float = 1e-6,
|
| 560 |
+
device: Union[str, torch.device] = None,
|
| 561 |
+
):
|
| 562 |
+
super().__init__()
|
| 563 |
+
self.weight = nn.Parameter(torch.ones(size, device=device))
|
| 564 |
+
self.eps = eps
|
| 565 |
+
|
| 566 |
+
def forward(self, x: torch.Tensor) -> torch.Tensor:
|
| 567 |
+
with torch.autocast(enabled=False, device_type=x.device.type):
|
| 568 |
+
og_dtype = x.dtype
|
| 569 |
+
x = x.to(torch.float32)
|
| 570 |
+
variance = x.pow(2).mean(-1, keepdim=True)
|
| 571 |
+
x = x * torch.rsqrt(variance + self.eps)
|
| 572 |
+
x = x.to(og_dtype)
|
| 573 |
+
|
| 574 |
+
return self.weight * x
|
| 575 |
+
|
| 576 |
+
def extra_repr(self):
|
| 577 |
+
return f"{tuple(self.weight.shape)}, eps={self.eps}"
|
| 578 |
+
|
| 579 |
+
|
| 580 |
+
# Copied from transformers.models.llama.modeling_llama.repeat_kv
|
| 581 |
+
def repeat_kv(hidden_states: torch.Tensor, n_rep: int) -> torch.Tensor:
|
| 582 |
+
"""
|
| 583 |
+
This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch,
|
| 584 |
+
num_key_value_heads, seqlen, head_dim) to (batch, num_attention_heads, seqlen, head_dim)
|
| 585 |
+
"""
|
| 586 |
+
batch, num_key_value_heads, slen, head_dim = hidden_states.shape
|
| 587 |
+
if n_rep == 1:
|
| 588 |
+
return hidden_states
|
| 589 |
+
hidden_states = hidden_states[:, :, None, :, :].expand(batch, num_key_value_heads, n_rep, slen, head_dim)
|
| 590 |
+
return hidden_states.reshape(batch, num_key_value_heads * n_rep, slen, head_dim)
|
| 591 |
+
|
| 592 |
+
|
| 593 |
+
def eager_attention_forward(
|
| 594 |
+
module: nn.Module,
|
| 595 |
+
query: torch.Tensor,
|
| 596 |
+
key: torch.Tensor,
|
| 597 |
+
value: torch.Tensor,
|
| 598 |
+
attention_mask: Optional[torch.Tensor],
|
| 599 |
+
scaling: float,
|
| 600 |
+
dropout: float = 0.0,
|
| 601 |
+
**kwargs,
|
| 602 |
+
) -> tuple[torch.Tensor, Optional[torch.Tensor]]:
|
| 603 |
+
key_states = repeat_kv(key, module.num_key_value_groups)
|
| 604 |
+
value_states = repeat_kv(value, module.num_key_value_groups)
|
| 605 |
+
|
| 606 |
+
attn_weights = torch.matmul(query, key_states.transpose(2, 3)) * scaling
|
| 607 |
+
if attention_mask is not None:
|
| 608 |
+
causal_mask = attention_mask[:, :, :, : key_states.shape[-2]]
|
| 609 |
+
attn_weights = attn_weights + causal_mask
|
| 610 |
+
|
| 611 |
+
attn_weights = nn.functional.softmax(attn_weights, dim=-1, dtype=torch.float32).to(query.dtype)
|
| 612 |
+
attn_weights = nn.functional.dropout(attn_weights, p=dropout, training=module.training)
|
| 613 |
+
attn_output = torch.matmul(attn_weights, value_states)
|
| 614 |
+
attn_output = attn_output.transpose(1, 2).contiguous()
|
| 615 |
+
|
| 616 |
+
return attn_output, attn_weights
|
| 617 |
+
|
| 618 |
+
|
| 619 |
+
class Molmo2Attention(nn.Module):
|
| 620 |
+
"""Multi-headed attention from 'Attention Is All You Need' paper"""
|
| 621 |
+
|
| 622 |
+
def __init__(self, config: Molmo2TextConfig, layer_idx: int) -> None:
|
| 623 |
+
super().__init__()
|
| 624 |
+
self.config = config
|
| 625 |
+
self.layer_idx = layer_idx
|
| 626 |
+
self.num_heads = config.num_attention_heads
|
| 627 |
+
self.num_key_value_heads = config.num_key_value_heads
|
| 628 |
+
self.num_key_value_groups = config.num_attention_heads // config.num_key_value_heads
|
| 629 |
+
self.head_dim = config.head_dim
|
| 630 |
+
self.scaling = self.head_dim**-0.5
|
| 631 |
+
self.is_causal = True
|
| 632 |
+
|
| 633 |
+
self.fused_dims = (
|
| 634 |
+
config.num_attention_heads * config.head_dim,
|
| 635 |
+
config.head_dim * config.num_key_value_heads,
|
| 636 |
+
config.head_dim * config.num_key_value_heads,
|
| 637 |
+
)
|
| 638 |
+
self.att_proj = nn.Linear(
|
| 639 |
+
config.hidden_size,
|
| 640 |
+
sum(self.fused_dims),
|
| 641 |
+
bias=config.qkv_bias,
|
| 642 |
+
)
|
| 643 |
+
|
| 644 |
+
# Layer norms.
|
| 645 |
+
self.k_norm: Optional[Molmo2RMSNorm] = None
|
| 646 |
+
self.q_norm: Optional[Molmo2RMSNorm] = None
|
| 647 |
+
self.qk_norm_type: Optional[str] = None
|
| 648 |
+
if config.use_qk_norm:
|
| 649 |
+
k_norm_size = (
|
| 650 |
+
config.head_dim
|
| 651 |
+
if config.qk_norm_type == "qwen3" else
|
| 652 |
+
config.num_key_value_heads * config.head_dim
|
| 653 |
+
)
|
| 654 |
+
self.k_norm = Molmo2RMSNorm(k_norm_size, eps=config.layer_norm_eps)
|
| 655 |
+
q_norm_size = (
|
| 656 |
+
config.head_dim
|
| 657 |
+
if config.qk_norm_type == "qwen3" else
|
| 658 |
+
config.num_attention_heads * config.head_dim
|
| 659 |
+
)
|
| 660 |
+
self.q_norm = Molmo2RMSNorm(q_norm_size, eps=config.layer_norm_eps)
|
| 661 |
+
self.qk_norm_type = config.qk_norm_type
|
| 662 |
+
|
| 663 |
+
self.attention_dropout = config.attention_dropout
|
| 664 |
+
|
| 665 |
+
self.attn_out = nn.Linear(
|
| 666 |
+
config.head_dim * config.num_attention_heads,
|
| 667 |
+
config.hidden_size,
|
| 668 |
+
bias=False,
|
| 669 |
+
)
|
| 670 |
+
|
| 671 |
+
def forward(
|
| 672 |
+
self,
|
| 673 |
+
hidden_states: torch.Tensor,
|
| 674 |
+
position_embeddings: tuple[torch.Tensor, torch.Tensor],
|
| 675 |
+
attention_mask: Optional[torch.Tensor],
|
| 676 |
+
past_key_values: Optional[Cache] = None,
|
| 677 |
+
cache_position: Optional[torch.LongTensor] = None,
|
| 678 |
+
**kwargs: Unpack[FlashAttentionKwargs],
|
| 679 |
+
) -> tuple[torch.Tensor, Optional[torch.Tensor], Optional[tuple[torch.Tensor]]]:
|
| 680 |
+
input_shape = hidden_states.shape[:-1]
|
| 681 |
+
hidden_shape = (*input_shape, -1, self.head_dim)
|
| 682 |
+
|
| 683 |
+
qkv = self.att_proj(hidden_states)
|
| 684 |
+
query_states, key_states, value_states = qkv.split(self.fused_dims, dim=-1)
|
| 685 |
+
value_states = value_states.view(hidden_shape)
|
| 686 |
+
|
| 687 |
+
# Optionally apply layer norm to keys and queries.
|
| 688 |
+
if self.q_norm is not None and self.k_norm is not None and self.qk_norm_type != "qwen3":
|
| 689 |
+
query_states = self.q_norm(query_states)
|
| 690 |
+
key_states = self.k_norm(key_states)
|
| 691 |
+
|
| 692 |
+
query_states = query_states.view(hidden_shape)
|
| 693 |
+
key_states = key_states.view(hidden_shape)
|
| 694 |
+
if self.q_norm is not None and self.k_norm is not None and self.qk_norm_type == "qwen3":
|
| 695 |
+
query_states = self.q_norm(query_states)
|
| 696 |
+
key_states = self.k_norm(key_states)
|
| 697 |
+
query_states = query_states.transpose(1, 2)
|
| 698 |
+
key_states = key_states.transpose(1, 2)
|
| 699 |
+
value_states = value_states.transpose(1, 2)
|
| 700 |
+
|
| 701 |
+
cos, sin = position_embeddings
|
| 702 |
+
query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)
|
| 703 |
+
|
| 704 |
+
if past_key_values is not None:
|
| 705 |
+
# sin and cos are specific to RoPE models; cache_position needed for the static cache
|
| 706 |
+
cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}
|
| 707 |
+
key_states, value_states = past_key_values.update(key_states, value_states, self.layer_idx, cache_kwargs)
|
| 708 |
+
|
| 709 |
+
attention_interface: Callable = eager_attention_forward
|
| 710 |
+
if self.config._attn_implementation != "eager":
|
| 711 |
+
attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]
|
| 712 |
+
|
| 713 |
+
attn_output, attn_weights = attention_interface(
|
| 714 |
+
self,
|
| 715 |
+
query_states,
|
| 716 |
+
key_states,
|
| 717 |
+
value_states,
|
| 718 |
+
attention_mask,
|
| 719 |
+
dropout=0.0 if not self.training else self.attention_dropout,
|
| 720 |
+
scaling=self.scaling,
|
| 721 |
+
**kwargs,
|
| 722 |
+
)
|
| 723 |
+
|
| 724 |
+
attn_output = attn_output.reshape(*input_shape, -1).contiguous()
|
| 725 |
+
attn_output = self.attn_out(attn_output)
|
| 726 |
+
return attn_output, attn_weights
|
| 727 |
+
|
| 728 |
+
|
| 729 |
+
class LanguageModelMLP(nn.Module):
|
| 730 |
+
|
| 731 |
+
def __init__(
|
| 732 |
+
self,
|
| 733 |
+
input_dim: int,
|
| 734 |
+
intermediate_size: int,
|
| 735 |
+
hidden_act: str,
|
| 736 |
+
device: Union[str, torch.device] = None,
|
| 737 |
+
):
|
| 738 |
+
super().__init__()
|
| 739 |
+
self.ff_proj = nn.Linear(input_dim, intermediate_size * 2, bias=False, device=device)
|
| 740 |
+
self.ff_out = nn.Linear(intermediate_size, input_dim, bias=False, device=device)
|
| 741 |
+
self.act = ACT2FN[hidden_act]
|
| 742 |
+
|
| 743 |
+
def forward(self, x: torch.Tensor) -> torch.Tensor:
|
| 744 |
+
x = self.ff_proj(x)
|
| 745 |
+
x, gate = x.chunk(2, dim=-1)
|
| 746 |
+
x = self.act(gate) * x
|
| 747 |
+
x = self.ff_out(x)
|
| 748 |
+
return x
|
| 749 |
+
|
| 750 |
+
|
| 751 |
+
class Molmo2DecoderLayer(GradientCheckpointingLayer):
|
| 752 |
+
|
| 753 |
+
def __init__(
|
| 754 |
+
self,
|
| 755 |
+
config: Molmo2TextConfig,
|
| 756 |
+
layer_idx: Optional[int] = None,
|
| 757 |
+
device: Union[str, torch.device] = None
|
| 758 |
+
):
|
| 759 |
+
super().__init__()
|
| 760 |
+
self.config = config
|
| 761 |
+
|
| 762 |
+
self.self_attn = Molmo2Attention(config, layer_idx)
|
| 763 |
+
self.attn_norm = Molmo2RMSNorm(
|
| 764 |
+
config.hidden_size, eps=config.layer_norm_eps, device=device)
|
| 765 |
+
self.dropout = nn.Dropout(config.residual_dropout)
|
| 766 |
+
self.mlp = LanguageModelMLP(
|
| 767 |
+
config.hidden_size, config.intermediate_size, config.hidden_act, device=device)
|
| 768 |
+
self.ff_norm = Molmo2RMSNorm(
|
| 769 |
+
config.hidden_size, eps=config.layer_norm_eps, device=device)
|
| 770 |
+
|
| 771 |
+
def forward(
|
| 772 |
+
self,
|
| 773 |
+
hidden_states: torch.Tensor,
|
| 774 |
+
position_embeddings: tuple[torch.Tensor, torch.Tensor],
|
| 775 |
+
attention_mask: Optional[torch.Tensor] = None,
|
| 776 |
+
position_ids: Optional[torch.LongTensor] = None,
|
| 777 |
+
past_key_values: Optional[Cache] = None,
|
| 778 |
+
output_attentions: Optional[bool] = False,
|
| 779 |
+
use_cache: Optional[bool] = False,
|
| 780 |
+
cache_position: Optional[torch.LongTensor] = None,
|
| 781 |
+
**kwargs: Unpack[TransformersKwargs],
|
| 782 |
+
) -> tuple[torch.FloatTensor, Optional[tuple[torch.FloatTensor, torch.FloatTensor]]]:
|
| 783 |
+
|
| 784 |
+
residual = hidden_states
|
| 785 |
+
hidden_states = self.attn_norm(hidden_states)
|
| 786 |
+
|
| 787 |
+
# Self Attention
|
| 788 |
+
hidden_states, self_attn_weights = self.self_attn(
|
| 789 |
+
hidden_states=hidden_states,
|
| 790 |
+
position_embeddings=position_embeddings,
|
| 791 |
+
attention_mask=attention_mask,
|
| 792 |
+
position_ids=position_ids,
|
| 793 |
+
past_key_values=past_key_values,
|
| 794 |
+
output_attentions=output_attentions,
|
| 795 |
+
use_cache=use_cache,
|
| 796 |
+
cache_position=cache_position,
|
| 797 |
+
**kwargs,
|
| 798 |
+
)
|
| 799 |
+
|
| 800 |
+
hidden_states = residual + self.dropout(hidden_states)
|
| 801 |
+
|
| 802 |
+
# Fully Connected
|
| 803 |
+
residual = hidden_states
|
| 804 |
+
hidden_states = self.ff_norm(hidden_states)
|
| 805 |
+
hidden_states = self.mlp(hidden_states)
|
| 806 |
+
|
| 807 |
+
hidden_states = residual + self.dropout(hidden_states)
|
| 808 |
+
|
| 809 |
+
outputs = (hidden_states,)
|
| 810 |
+
|
| 811 |
+
if output_attentions:
|
| 812 |
+
outputs += (self_attn_weights,)
|
| 813 |
+
|
| 814 |
+
return outputs
|
| 815 |
+
|
| 816 |
+
|
| 817 |
+
class Molmo2PostNormDecoderLayer(Molmo2DecoderLayer):
|
| 818 |
+
def forward(
|
| 819 |
+
self,
|
| 820 |
+
hidden_states: torch.Tensor,
|
| 821 |
+
position_embeddings: tuple[torch.Tensor, torch.Tensor],
|
| 822 |
+
attention_mask: Optional[torch.Tensor] = None,
|
| 823 |
+
position_ids: Optional[torch.LongTensor] = None,
|
| 824 |
+
past_key_values: Optional[Cache] = None,
|
| 825 |
+
output_attentions: Optional[bool] = False,
|
| 826 |
+
use_cache: Optional[bool] = False,
|
| 827 |
+
cache_position: Optional[torch.LongTensor] = None,
|
| 828 |
+
**kwargs,
|
| 829 |
+
) -> tuple[torch.FloatTensor, Optional[tuple[torch.FloatTensor, torch.FloatTensor]]]:
|
| 830 |
+
|
| 831 |
+
residual = hidden_states
|
| 832 |
+
|
| 833 |
+
# Self Attention
|
| 834 |
+
hidden_states, self_attn_weights = self.self_attn(
|
| 835 |
+
hidden_states=hidden_states,
|
| 836 |
+
position_embeddings=position_embeddings,
|
| 837 |
+
attention_mask=attention_mask,
|
| 838 |
+
position_ids=position_ids,
|
| 839 |
+
past_key_values=past_key_values,
|
| 840 |
+
output_attentions=output_attentions,
|
| 841 |
+
use_cache=use_cache,
|
| 842 |
+
cache_position=cache_position,
|
| 843 |
+
)
|
| 844 |
+
hidden_states = self.attn_norm(hidden_states)
|
| 845 |
+
|
| 846 |
+
hidden_states = residual + self.dropout(hidden_states)
|
| 847 |
+
|
| 848 |
+
# Fully Connected
|
| 849 |
+
residual = hidden_states
|
| 850 |
+
hidden_states = self.mlp(hidden_states)
|
| 851 |
+
hidden_states = self.ff_norm(hidden_states)
|
| 852 |
+
|
| 853 |
+
hidden_states = residual + self.dropout(hidden_states)
|
| 854 |
+
|
| 855 |
+
outputs = (hidden_states,)
|
| 856 |
+
|
| 857 |
+
if output_attentions:
|
| 858 |
+
outputs += (self_attn_weights,)
|
| 859 |
+
|
| 860 |
+
return outputs
|
| 861 |
+
|
| 862 |
+
|
| 863 |
+
class Molmo2Embedding(nn.Module):
|
| 864 |
+
def __init__(
|
| 865 |
+
self,
|
| 866 |
+
num_embeddings: int,
|
| 867 |
+
num_new_embeddings: int,
|
| 868 |
+
features: int,
|
| 869 |
+
device: Union[str, torch.device] = None,
|
| 870 |
+
):
|
| 871 |
+
super().__init__()
|
| 872 |
+
self.embedding = nn.Parameter(
|
| 873 |
+
torch.zeros(num_embeddings, features, device=device),
|
| 874 |
+
)
|
| 875 |
+
self.new_embedding = nn.Parameter(
|
| 876 |
+
torch.zeros(num_new_embeddings, features, device=device),
|
| 877 |
+
)
|
| 878 |
+
|
| 879 |
+
def forward(self, x: torch.Tensor) -> torch.Tensor:
|
| 880 |
+
return F.embedding(x, torch.cat([self.embedding, self.new_embedding], dim=0))
|
| 881 |
+
|
| 882 |
+
|
| 883 |
+
class Molmo2PreTrainedModel(PreTrainedModel):
|
| 884 |
+
config: Molmo2Config
|
| 885 |
+
base_model_prefix = "model"
|
| 886 |
+
supports_gradient_checkpointing = True
|
| 887 |
+
_no_split_modules = [
|
| 888 |
+
"Molmo2DecoderLayer",
|
| 889 |
+
"Molmo2PostNormDecoderLayer",
|
| 890 |
+
"Molmo2VisionBlock",
|
| 891 |
+
"ViTMultiHeadDotProductAttention",
|
| 892 |
+
]
|
| 893 |
+
_skip_keys_device_placement = "past_key_values"
|
| 894 |
+
_supports_flash_attn = True
|
| 895 |
+
_supports_sdpa = True
|
| 896 |
+
|
| 897 |
+
_can_compile_fullgraph = True
|
| 898 |
+
_supports_attention_backend = True
|
| 899 |
+
_can_record_outputs = {
|
| 900 |
+
"hidden_states": Molmo2DecoderLayer,
|
| 901 |
+
"attentions": Molmo2Attention,
|
| 902 |
+
}
|
| 903 |
+
|
| 904 |
+
def _init_weights(self, module):
|
| 905 |
+
std = self.config.initializer_range
|
| 906 |
+
if isinstance(module, (nn.Linear,)):
|
| 907 |
+
module.weight.data.normal_(mean=0.0, std=std)
|
| 908 |
+
if module.bias is not None:
|
| 909 |
+
module.bias.data.zero_()
|
| 910 |
+
elif isinstance(module, Molmo2Embedding):
|
| 911 |
+
module.embedding.data.normal_(mean=0.0, std=std)
|
| 912 |
+
module.new_embedding.data.normal_(mean=0.0, std=std)
|
| 913 |
+
elif isinstance(module, nn.Embedding):
|
| 914 |
+
module.weight.data.normal_(mean=0.0, std=std)
|
| 915 |
+
if module.padding_idx is not None:
|
| 916 |
+
module.weight.data[module.padding_idx].zero_()
|
| 917 |
+
elif isinstance(module, Molmo2RMSNorm):
|
| 918 |
+
module.weight.data.fill_(1.0)
|
| 919 |
+
elif isinstance(module, nn.LayerNorm):
|
| 920 |
+
module.weight.data.fill_(1.0)
|
| 921 |
+
if module.bias is not None:
|
| 922 |
+
module.bias.data.zero_()
|
| 923 |
+
|
| 924 |
+
|
| 925 |
+
class Molmo2TextModel(Molmo2PreTrainedModel):
|
| 926 |
+
config: Molmo2TextConfig
|
| 927 |
+
_no_split_modules = ["Molmo2DecoderLayer", "Molmo2PostNormDecoderLayer"]
|
| 928 |
+
|
| 929 |
+
def __init__(self, config: Molmo2TextConfig):
|
| 930 |
+
super().__init__(config)
|
| 931 |
+
if config.additional_vocab_size is not None:
|
| 932 |
+
self.wte = Molmo2Embedding(
|
| 933 |
+
config.vocab_size,
|
| 934 |
+
config.additional_vocab_size,
|
| 935 |
+
config.hidden_size,
|
| 936 |
+
)
|
| 937 |
+
else:
|
| 938 |
+
self.wte = nn.Embedding(config.vocab_size, config.hidden_size)
|
| 939 |
+
self.emb_drop = nn.Dropout(config.embedding_dropout)
|
| 940 |
+
decoder_layer = Molmo2PostNormDecoderLayer if config.norm_after else Molmo2DecoderLayer
|
| 941 |
+
self.blocks = nn.ModuleList(
|
| 942 |
+
[decoder_layer(config, layer_idx) for layer_idx in range(config.num_hidden_layers)]
|
| 943 |
+
)
|
| 944 |
+
self.ln_f = Molmo2RMSNorm(config.hidden_size, eps=config.layer_norm_eps)
|
| 945 |
+
if config.rope_scaling_layers is not None:
|
| 946 |
+
self.rotary_embs = nn.ModuleDict(
|
| 947 |
+
{
|
| 948 |
+
"default": Molmo2RotaryEmbedding(config, rope_type="default"),
|
| 949 |
+
"scaling": Molmo2RotaryEmbedding(config),
|
| 950 |
+
}
|
| 951 |
+
)
|
| 952 |
+
else:
|
| 953 |
+
self.rotary_emb = Molmo2RotaryEmbedding(config)
|
| 954 |
+
self.gradient_checkpointing = False
|
| 955 |
+
|
| 956 |
+
# Initialize weights and apply final processing
|
| 957 |
+
self.post_init()
|
| 958 |
+
|
| 959 |
+
def get_input_embeddings(self) -> torch.nn.Module:
|
| 960 |
+
return self.wte
|
| 961 |
+
|
| 962 |
+
def set_input_embeddings(self, value: torch.nn.Module) -> None:
|
| 963 |
+
self.wte = value
|
| 964 |
+
|
| 965 |
+
@can_return_tuple
|
| 966 |
+
def forward(
|
| 967 |
+
self,
|
| 968 |
+
input_ids: Optional[torch.LongTensor] = None,
|
| 969 |
+
attention_mask: Optional[torch.Tensor] = None,
|
| 970 |
+
position_ids: Optional[torch.LongTensor] = None,
|
| 971 |
+
past_key_values: Optional[Cache] = None,
|
| 972 |
+
inputs_embeds: Optional[torch.FloatTensor] = None,
|
| 973 |
+
use_cache: Optional[bool] = None,
|
| 974 |
+
output_attentions: Optional[bool] = None,
|
| 975 |
+
output_hidden_states: Optional[bool] = None,
|
| 976 |
+
cache_position: Optional[torch.LongTensor] = None,
|
| 977 |
+
**kwargs: Unpack[TransformersKwargs],
|
| 978 |
+
) -> BaseModelOutputWithPast:
|
| 979 |
+
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
|
| 980 |
+
output_hidden_states = (
|
| 981 |
+
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
|
| 982 |
+
)
|
| 983 |
+
use_cache = use_cache if use_cache is not None else self.config.use_cache
|
| 984 |
+
|
| 985 |
+
if (input_ids is None) ^ (inputs_embeds is not None):
|
| 986 |
+
raise ValueError("You must specify exactly one of input_ids or inputs_embeds")
|
| 987 |
+
|
| 988 |
+
if self.gradient_checkpointing and self.training and use_cache:
|
| 989 |
+
logger.warning_once(
|
| 990 |
+
"`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`."
|
| 991 |
+
)
|
| 992 |
+
use_cache = False
|
| 993 |
+
|
| 994 |
+
if inputs_embeds is None:
|
| 995 |
+
input_ids = input_ids * (input_ids != -1).to(input_ids.dtype)
|
| 996 |
+
inputs_embeds = self.wte(input_ids)
|
| 997 |
+
|
| 998 |
+
# torch.jit.trace() doesn't support cache objects in the output
|
| 999 |
+
if use_cache and past_key_values is None and not torch.jit.is_tracing():
|
| 1000 |
+
past_key_values = DynamicCache(config=self.config)
|
| 1001 |
+
|
| 1002 |
+
if cache_position is None:
|
| 1003 |
+
past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
|
| 1004 |
+
cache_position = torch.arange(
|
| 1005 |
+
past_seen_tokens,
|
| 1006 |
+
past_seen_tokens + inputs_embeds.shape[1],
|
| 1007 |
+
device=inputs_embeds.device,
|
| 1008 |
+
)
|
| 1009 |
+
|
| 1010 |
+
if position_ids is None:
|
| 1011 |
+
position_ids = cache_position.unsqueeze(0)
|
| 1012 |
+
|
| 1013 |
+
# It may already have been prepared by e.g. `generate`
|
| 1014 |
+
if not isinstance(causal_mask_mapping := attention_mask, dict):
|
| 1015 |
+
# Prepare mask arguments
|
| 1016 |
+
mask_kwargs = {
|
| 1017 |
+
"config": self.config,
|
| 1018 |
+
"input_embeds": inputs_embeds,
|
| 1019 |
+
"attention_mask": attention_mask,
|
| 1020 |
+
"cache_position": cache_position,
|
| 1021 |
+
"past_key_values": past_key_values,
|
| 1022 |
+
"position_ids": position_ids,
|
| 1023 |
+
}
|
| 1024 |
+
|
| 1025 |
+
# Create the mask
|
| 1026 |
+
causal_mask_mapping = create_causal_mask(**mask_kwargs)
|
| 1027 |
+
|
| 1028 |
+
hidden_states = inputs_embeds
|
| 1029 |
+
|
| 1030 |
+
# create position embeddings to be shared across the decoder layers
|
| 1031 |
+
if self.config.rope_scaling_layers is not None:
|
| 1032 |
+
position_embeddings_mapping = {
|
| 1033 |
+
"default": self.rotary_embs["default"](hidden_states, position_ids),
|
| 1034 |
+
"scaling": self.rotary_embs["scaling"](hidden_states, position_ids),
|
| 1035 |
+
}
|
| 1036 |
+
else:
|
| 1037 |
+
position_embeddings = self.rotary_emb(hidden_states, position_ids)
|
| 1038 |
+
|
| 1039 |
+
# decoder layers
|
| 1040 |
+
all_hidden_states = () if output_hidden_states else None
|
| 1041 |
+
all_self_attns = () if output_attentions else None
|
| 1042 |
+
|
| 1043 |
+
for layer_idx, decoder_block in enumerate(self.blocks[: self.config.num_hidden_layers]):
|
| 1044 |
+
if output_hidden_states:
|
| 1045 |
+
all_hidden_states += (hidden_states,)
|
| 1046 |
+
|
| 1047 |
+
if self.config.rope_scaling_layers is not None:
|
| 1048 |
+
position_embeddings_i = (
|
| 1049 |
+
position_embeddings_mapping["scaling"]
|
| 1050 |
+
if layer_idx in self.config.rope_scaling_layers
|
| 1051 |
+
else position_embeddings_mapping["default"]
|
| 1052 |
+
)
|
| 1053 |
+
else:
|
| 1054 |
+
position_embeddings_i = position_embeddings
|
| 1055 |
+
|
| 1056 |
+
layer_outputs = decoder_block(
|
| 1057 |
+
hidden_states,
|
| 1058 |
+
attention_mask=causal_mask_mapping,
|
| 1059 |
+
position_ids=position_ids,
|
| 1060 |
+
past_key_values=past_key_values,
|
| 1061 |
+
output_attentions=output_attentions,
|
| 1062 |
+
use_cache=use_cache,
|
| 1063 |
+
cache_position=cache_position,
|
| 1064 |
+
position_embeddings=position_embeddings_i,
|
| 1065 |
+
**kwargs,
|
| 1066 |
+
)
|
| 1067 |
+
|
| 1068 |
+
hidden_states = layer_outputs[0]
|
| 1069 |
+
|
| 1070 |
+
if output_attentions:
|
| 1071 |
+
all_self_attns += (layer_outputs[1],)
|
| 1072 |
+
|
| 1073 |
+
hidden_states = self.ln_f(hidden_states)
|
| 1074 |
+
|
| 1075 |
+
# add hidden states from the last decoder layer
|
| 1076 |
+
if output_hidden_states:
|
| 1077 |
+
all_hidden_states += (hidden_states,)
|
| 1078 |
+
|
| 1079 |
+
return BaseModelOutputWithPast(
|
| 1080 |
+
last_hidden_state=hidden_states,
|
| 1081 |
+
past_key_values=past_key_values,
|
| 1082 |
+
hidden_states=all_hidden_states,
|
| 1083 |
+
attentions=all_self_attns,
|
| 1084 |
+
)
|
| 1085 |
+
|
| 1086 |
+
# Adapted from transformers.models.gemma3.modeling_gemma3
|
| 1087 |
+
def token_type_ids_mask_function(
|
| 1088 |
+
token_type_ids: Optional[torch.Tensor] = None,
|
| 1089 |
+
) -> Optional[Callable]:
|
| 1090 |
+
"""
|
| 1091 |
+
This function adds the correct offsets to the `q_idx` and `kv_idx` as the torch API can only accept lengths,
|
| 1092 |
+
not start and end indices.
|
| 1093 |
+
"""
|
| 1094 |
+
# Do not return an additional mask in this case
|
| 1095 |
+
if token_type_ids is None:
|
| 1096 |
+
return None
|
| 1097 |
+
|
| 1098 |
+
def inner_mask(batch_idx: int, head_idx: int, q_idx: int, kv_idx: int) -> bool:
|
| 1099 |
+
# If it's 1 for both query and key/value, we are in an image block
|
| 1100 |
+
# NOTE: static cache shape goes beyond input seq length, while token_type_ids.shape[1] == input seq length
|
| 1101 |
+
# Since vmap doesn't support `if statement` we workaround it with `torch.where`
|
| 1102 |
+
safe_idx = torch.where(kv_idx < token_type_ids.shape[1], kv_idx, 0)
|
| 1103 |
+
token_type_ids_at_kv_idx = token_type_ids[batch_idx, safe_idx]
|
| 1104 |
+
token_type_ids_at_kv_idx = torch.where(kv_idx < token_type_ids.shape[1], token_type_ids_at_kv_idx, 0)
|
| 1105 |
+
|
| 1106 |
+
is_image_block = (token_type_ids[batch_idx, q_idx] == 1) & (token_type_ids_at_kv_idx == 1)
|
| 1107 |
+
|
| 1108 |
+
# This is bidirectional attention whenever we are dealing with image tokens
|
| 1109 |
+
return is_image_block & is_image_block
|
| 1110 |
+
|
| 1111 |
+
return inner_mask
|
| 1112 |
+
|
| 1113 |
+
|
| 1114 |
+
class Molmo2Model(Molmo2PreTrainedModel):
|
| 1115 |
+
base_model_prefix = ""
|
| 1116 |
+
_checkpoint_conversion_mapping = {}
|
| 1117 |
+
# Reference: fix gemma3 grad acc #37208
|
| 1118 |
+
accepts_loss_kwargs = False
|
| 1119 |
+
config: Molmo2Config
|
| 1120 |
+
|
| 1121 |
+
|
| 1122 |
+
def __init__(self, config: Molmo2Config):
|
| 1123 |
+
super().__init__(config)
|
| 1124 |
+
self.transformer: Molmo2TextModel = Molmo2TextModel(config.text_config)
|
| 1125 |
+
self.vision_backbone: Optional[Molmo2VisionBackbone] = None
|
| 1126 |
+
if config.vit_config is not None and config.adapter_config is not None:
|
| 1127 |
+
self.vision_backbone = Molmo2VisionBackbone(config.vit_config, config.adapter_config)
|
| 1128 |
+
|
| 1129 |
+
# Initialize weights and apply final processing
|
| 1130 |
+
self.post_init()
|
| 1131 |
+
|
| 1132 |
+
def get_input_embeddings(self) -> torch.nn.Module:
|
| 1133 |
+
return self.transformer.wte
|
| 1134 |
+
|
| 1135 |
+
def set_input_embeddings(self, value: torch.nn.Module) -> None:
|
| 1136 |
+
self.transformer.wte = value
|
| 1137 |
+
|
| 1138 |
+
def set_decoder(self, decoder):
|
| 1139 |
+
self.transformer = decoder
|
| 1140 |
+
|
| 1141 |
+
def get_decoder(self):
|
| 1142 |
+
return self.transformer
|
| 1143 |
+
|
| 1144 |
+
@property
|
| 1145 |
+
def device(self) -> torch.device:
|
| 1146 |
+
return self.transformer.ln_f.weight.device
|
| 1147 |
+
|
| 1148 |
+
def build_batched_images(
|
| 1149 |
+
self,
|
| 1150 |
+
input_ids: torch.LongTensor,
|
| 1151 |
+
pixel_values: torch.Tensor,
|
| 1152 |
+
image_token_pooling: torch.Tensor,
|
| 1153 |
+
image_grids: torch.Tensor,
|
| 1154 |
+
image_num_crops: torch.Tensor,
|
| 1155 |
+
) -> tuple[torch.Tensor, torch.Tensor]:
|
| 1156 |
+
# 1) Count the number of images in each example
|
| 1157 |
+
raw_counts = (input_ids == self.config.image_end_token_id).sum(1) # [N]
|
| 1158 |
+
# Each image is represented by global view and high-res view
|
| 1159 |
+
# so we divide by 2 to get the number of images
|
| 1160 |
+
counts = raw_counts // 2
|
| 1161 |
+
N = counts.size(0)
|
| 1162 |
+
device = input_ids.device
|
| 1163 |
+
|
| 1164 |
+
# Total number of images in the batch
|
| 1165 |
+
num_images = int(counts.sum().item())
|
| 1166 |
+
|
| 1167 |
+
# Sanity check
|
| 1168 |
+
assert image_grids.size(0) == num_images, \
|
| 1169 |
+
f"Expected {num_images} image grids, but got {image_grids.size(0)}"
|
| 1170 |
+
assert image_num_crops.size(0) == num_images, \
|
| 1171 |
+
f"Expected {num_images} image num crops, but got {image_num_crops.size(0)}"
|
| 1172 |
+
|
| 1173 |
+
# 1-1) Compute per-image pooled patch count from image grids
|
| 1174 |
+
with torch.no_grad():
|
| 1175 |
+
first_prod = image_grids[:, :2].prod(dim=1) # [num_images]
|
| 1176 |
+
second_prod = image_grids[:, 2:].prod(dim=1) # [num_images]
|
| 1177 |
+
num_pooled_patches_per_image = (first_prod + second_prod).to(image_num_crops.dtype) # [num_images]
|
| 1178 |
+
|
| 1179 |
+
# pixel_values: [n_crops, n_patches, pixels_per_patch]
|
| 1180 |
+
n_crops, n_patches, pixels_per_patch = pixel_values.shape
|
| 1181 |
+
|
| 1182 |
+
# 2) Map each image index → example index
|
| 1183 |
+
# Example: if counts = [2, 1, 3], then this becomes [0,0,1,2,2,2]
|
| 1184 |
+
example_ids_for_image = torch.arange(N, device=device).repeat_interleave(counts) # [num_images]
|
| 1185 |
+
assert example_ids_for_image.numel() == num_images
|
| 1186 |
+
|
| 1187 |
+
# 2-1) Compute crops_per_example by summing per-image crop counts
|
| 1188 |
+
crops_per_example = torch.zeros(
|
| 1189 |
+
N, dtype=image_num_crops.dtype, device=image_num_crops.device
|
| 1190 |
+
)
|
| 1191 |
+
crops_per_example.index_add_(0, example_ids_for_image, image_num_crops) # [N]
|
| 1192 |
+
|
| 1193 |
+
# 2-2) Per-image number of patches = (crops per image) * n_patches
|
| 1194 |
+
patches_per_image = image_num_crops * n_patches # [num_images]
|
| 1195 |
+
|
| 1196 |
+
# 2-3) Compute per-example per-image patch offsets
|
| 1197 |
+
counts_list = counts.tolist()
|
| 1198 |
+
index_offset_per_example_list = []
|
| 1199 |
+
offset_img = 0
|
| 1200 |
+
for c in counts_list:
|
| 1201 |
+
per_img_patches = patches_per_image[offset_img:offset_img + c] # [c]
|
| 1202 |
+
# Offsets: [0, img0_total_patches, img0+img1_total_patches, ...]
|
| 1203 |
+
index_offset = [0] + per_img_patches.cumsum(0).tolist()[:-1]
|
| 1204 |
+
index_offset_per_example_list.append(index_offset)
|
| 1205 |
+
offset_img += c
|
| 1206 |
+
|
| 1207 |
+
# 2-4) Compute num_pooled_patches_per_example
|
| 1208 |
+
num_pooled_patches_per_example = torch.zeros(
|
| 1209 |
+
N, dtype=num_pooled_patches_per_image.dtype, device=num_pooled_patches_per_image.device
|
| 1210 |
+
)
|
| 1211 |
+
num_pooled_patches_per_example.index_add_(
|
| 1212 |
+
0, example_ids_for_image, num_pooled_patches_per_image
|
| 1213 |
+
)
|
| 1214 |
+
|
| 1215 |
+
# Sanity checks
|
| 1216 |
+
total_crops = int(crops_per_example.sum().item())
|
| 1217 |
+
assert total_crops == n_crops, \
|
| 1218 |
+
f"Expected {total_crops} crops, but got {n_crops}"
|
| 1219 |
+
|
| 1220 |
+
total_num_pooled_patches = int(num_pooled_patches_per_example.sum().item())
|
| 1221 |
+
assert total_num_pooled_patches == image_token_pooling.size(0), \
|
| 1222 |
+
f"Expected {total_num_pooled_patches} pooled patches, but got {image_token_pooling.size(0)}"
|
| 1223 |
+
|
| 1224 |
+
# 3) Build images tensor filled with -1
|
| 1225 |
+
M = int(crops_per_example.max().item())
|
| 1226 |
+
images = torch.full(
|
| 1227 |
+
(N, M, n_patches, pixels_per_patch),
|
| 1228 |
+
fill_value=-1,
|
| 1229 |
+
dtype=pixel_values.dtype,
|
| 1230 |
+
device=pixel_values.device,
|
| 1231 |
+
)
|
| 1232 |
+
|
| 1233 |
+
# 4) Fill images with per-example slices from pixel_values
|
| 1234 |
+
offset_crop = 0
|
| 1235 |
+
for i in range(N):
|
| 1236 |
+
num = int(crops_per_example[i].item())
|
| 1237 |
+
cur = pixel_values[offset_crop:offset_crop + num] # [num, n_patches, pixels_per_patch]
|
| 1238 |
+
images[i, :num] = cur
|
| 1239 |
+
offset_crop += num
|
| 1240 |
+
|
| 1241 |
+
# Sanity check
|
| 1242 |
+
assert offset_crop == n_crops
|
| 1243 |
+
|
| 1244 |
+
# 5) Build new_token_pooling tensor filled with -1
|
| 1245 |
+
P = int(num_pooled_patches_per_example.max().item())
|
| 1246 |
+
_, dim = image_token_pooling.shape
|
| 1247 |
+
new_token_pooling = torch.full(
|
| 1248 |
+
(N, P, dim),
|
| 1249 |
+
fill_value=-1,
|
| 1250 |
+
dtype=image_token_pooling.dtype,
|
| 1251 |
+
device=image_token_pooling.device,
|
| 1252 |
+
)
|
| 1253 |
+
|
| 1254 |
+
# 6) Fill token_pooling with per-example slices, adding per-image patch offsets
|
| 1255 |
+
patch_offset = 0
|
| 1256 |
+
img_offset = 0
|
| 1257 |
+
|
| 1258 |
+
for i, c in enumerate(counts_list):
|
| 1259 |
+
num_patches = int(num_pooled_patches_per_example[i].item())
|
| 1260 |
+
|
| 1261 |
+
# Subsequence of pooled tokens belonging to this example
|
| 1262 |
+
cur = image_token_pooling[patch_offset:patch_offset + num_patches].clone() # [num_patches, dim]
|
| 1263 |
+
|
| 1264 |
+
index_offset_per_example = index_offset_per_example_list[i] # length = c
|
| 1265 |
+
per_img_pooled = num_pooled_patches_per_image[img_offset:img_offset + c] # [c]
|
| 1266 |
+
|
| 1267 |
+
assert len(index_offset_per_example) == per_img_pooled.numel()
|
| 1268 |
+
|
| 1269 |
+
# Apply per-image offsets to the (ragged) subsequence
|
| 1270 |
+
offset = 0
|
| 1271 |
+
for j in range(c):
|
| 1272 |
+
index_offset = int(index_offset_per_example[j])
|
| 1273 |
+
n = int(per_img_pooled[j].item())
|
| 1274 |
+
cur_slice = cur[offset:offset + n]
|
| 1275 |
+
|
| 1276 |
+
# Apply offset across all columns
|
| 1277 |
+
cur[offset:offset + n] = torch.where(
|
| 1278 |
+
cur_slice >= 0,
|
| 1279 |
+
cur_slice + index_offset,
|
| 1280 |
+
cur_slice,
|
| 1281 |
+
)
|
| 1282 |
+
offset += n
|
| 1283 |
+
|
| 1284 |
+
new_token_pooling[i, :num_patches] = cur
|
| 1285 |
+
|
| 1286 |
+
patch_offset += num_patches
|
| 1287 |
+
img_offset += c
|
| 1288 |
+
|
| 1289 |
+
# Final sanity checks
|
| 1290 |
+
assert patch_offset == total_num_pooled_patches
|
| 1291 |
+
assert img_offset == num_images
|
| 1292 |
+
|
| 1293 |
+
return images, new_token_pooling
|
| 1294 |
+
|
| 1295 |
+
def build_batched_videos(
|
| 1296 |
+
self,
|
| 1297 |
+
input_ids: torch.LongTensor,
|
| 1298 |
+
pixel_values_videos: torch.Tensor,
|
| 1299 |
+
video_token_pooling: torch.Tensor,
|
| 1300 |
+
video_grids: torch.Tensor,
|
| 1301 |
+
) -> tuple[torch.Tensor, torch.Tensor]:
|
| 1302 |
+
|
| 1303 |
+
# 1) Count the number of videos in each example
|
| 1304 |
+
if self.config.use_frame_special_tokens:
|
| 1305 |
+
end_token_id = self.config.frame_end_token_id
|
| 1306 |
+
else:
|
| 1307 |
+
end_token_id = self.config.image_end_token_id
|
| 1308 |
+
counts = (input_ids == end_token_id).any(dim=1).long() # [N]
|
| 1309 |
+
N = counts.size(0)
|
| 1310 |
+
device = input_ids.device
|
| 1311 |
+
|
| 1312 |
+
# Total number of videos in the batch
|
| 1313 |
+
num_videos = int(counts.sum().item())
|
| 1314 |
+
|
| 1315 |
+
# Sanity check
|
| 1316 |
+
assert video_grids.size(0) == num_videos, \
|
| 1317 |
+
f"Expected {num_videos} videos, but got {video_grids.size(0)}"
|
| 1318 |
+
|
| 1319 |
+
video_num_frames = video_grids[:, 0] # [num_videos]
|
| 1320 |
+
num_pooled_patches_per_video = video_grids.prod(dim=1) # [num_videos]
|
| 1321 |
+
|
| 1322 |
+
# pixel_values_videos: [n_frames, n_patches, pixels_per_patch]
|
| 1323 |
+
n_frames, n_patches, pixels_per_patch = pixel_values_videos.shape
|
| 1324 |
+
|
| 1325 |
+
# 2) Map each video index -> example index
|
| 1326 |
+
# Example: if counts = [2, 1, 3], then this becomes [0,0,1,2,2,2]
|
| 1327 |
+
example_ids_for_video = torch.arange(N, device=device).repeat_interleave(counts) # [num_videos]
|
| 1328 |
+
assert example_ids_for_video.numel() == num_videos
|
| 1329 |
+
|
| 1330 |
+
# 2-1) Compute frames_per_example by summing per-video frame counts
|
| 1331 |
+
frames_per_example = torch.zeros(
|
| 1332 |
+
N, dtype=video_num_frames.dtype, device=device,
|
| 1333 |
+
)
|
| 1334 |
+
frames_per_example.index_add_(0, example_ids_for_video, video_num_frames) # [N]
|
| 1335 |
+
|
| 1336 |
+
# 2-2) Compute num_pooled_patches_per_example
|
| 1337 |
+
num_pooled_patches_per_example = torch.zeros(
|
| 1338 |
+
N, dtype=num_pooled_patches_per_video.dtype, device=num_pooled_patches_per_video.device,
|
| 1339 |
+
)
|
| 1340 |
+
num_pooled_patches_per_example.index_add_(
|
| 1341 |
+
0, example_ids_for_video, num_pooled_patches_per_video,
|
| 1342 |
+
)
|
| 1343 |
+
|
| 1344 |
+
# Sanity checks
|
| 1345 |
+
total_frames = int(frames_per_example.sum().item())
|
| 1346 |
+
assert total_frames == n_frames, \
|
| 1347 |
+
f"Expected {total_frames} frames, but got {n_frames}"
|
| 1348 |
+
|
| 1349 |
+
total_num_pooled_patches = int(num_pooled_patches_per_example.sum().item())
|
| 1350 |
+
assert total_num_pooled_patches == video_token_pooling.size(0), \
|
| 1351 |
+
f"Expected {total_num_pooled_patches} pooled patches, but got {video_token_pooling.size(0)}"
|
| 1352 |
+
|
| 1353 |
+
# 3) Build videos tensor filled with -1
|
| 1354 |
+
M = int(frames_per_example.max().item())
|
| 1355 |
+
videos = torch.full(
|
| 1356 |
+
(N, M, n_patches, pixels_per_patch),
|
| 1357 |
+
fill_value=-1,
|
| 1358 |
+
dtype=pixel_values_videos.dtype,
|
| 1359 |
+
device=device,
|
| 1360 |
+
)
|
| 1361 |
+
|
| 1362 |
+
# 4) Fill videos with per-examples slices from pixel_values_videos
|
| 1363 |
+
offset_frame = 0
|
| 1364 |
+
for i in range(N):
|
| 1365 |
+
num = int(frames_per_example[i].item())
|
| 1366 |
+
cur = pixel_values_videos[offset_frame:offset_frame + num] # [num, n_patches, pixels_per_patch]
|
| 1367 |
+
videos[i, :num] = cur
|
| 1368 |
+
offset_frame += num
|
| 1369 |
+
|
| 1370 |
+
# Sanity check
|
| 1371 |
+
assert offset_frame == n_frames
|
| 1372 |
+
|
| 1373 |
+
# 5) Build new token_pooling tensor filled with -1
|
| 1374 |
+
P = int(num_pooled_patches_per_example.max().item())
|
| 1375 |
+
_, dim = video_token_pooling.shape
|
| 1376 |
+
new_token_pooling = torch.full(
|
| 1377 |
+
(N, P, dim),
|
| 1378 |
+
fill_value=-1,
|
| 1379 |
+
dtype=video_token_pooling.dtype,
|
| 1380 |
+
device=video_token_pooling.device,
|
| 1381 |
+
)
|
| 1382 |
+
|
| 1383 |
+
# 6) Fill new token_pooling with per-examples slices from video_token_pooling
|
| 1384 |
+
patch_offset = 0
|
| 1385 |
+
for i in range(N):
|
| 1386 |
+
num_patches = int(num_pooled_patches_per_example[i].item())
|
| 1387 |
+
cur = video_token_pooling[patch_offset:patch_offset + num_patches] # [num_patches, dim]
|
| 1388 |
+
new_token_pooling[i, :num_patches] = cur
|
| 1389 |
+
patch_offset += num_patches
|
| 1390 |
+
|
| 1391 |
+
# Final sanity checks
|
| 1392 |
+
assert patch_offset == total_num_pooled_patches
|
| 1393 |
+
|
| 1394 |
+
return videos, new_token_pooling
|
| 1395 |
+
|
| 1396 |
+
def merge_visual_inputs(
|
| 1397 |
+
self,
|
| 1398 |
+
input_ids: Optional[torch.LongTensor] = None,
|
| 1399 |
+
pixel_values: Optional[torch.Tensor] = None,
|
| 1400 |
+
image_token_pooling: Optional[torch.Tensor] = None,
|
| 1401 |
+
image_grids: Optional[torch.Tensor] = None,
|
| 1402 |
+
image_num_crops: Optional[torch.Tensor] = None,
|
| 1403 |
+
pixel_values_videos: Optional[torch.Tensor] = None,
|
| 1404 |
+
video_token_pooling: Optional[torch.Tensor] = None,
|
| 1405 |
+
video_grids: Optional[torch.Tensor] = None,
|
| 1406 |
+
) -> tuple[Optional[torch.Tensor], Optional[torch.Tensor]]:
|
| 1407 |
+
if pixel_values is not None and pixel_values_videos is not None:
|
| 1408 |
+
raise ValueError("pixel_values and pixel_values_videos are provided at the same time")
|
| 1409 |
+
elif pixel_values is not None:
|
| 1410 |
+
assert input_ids is not None
|
| 1411 |
+
images, token_pooling = self.build_batched_images(
|
| 1412 |
+
input_ids=input_ids,
|
| 1413 |
+
pixel_values=pixel_values,
|
| 1414 |
+
image_token_pooling=image_token_pooling,
|
| 1415 |
+
image_grids=image_grids,
|
| 1416 |
+
image_num_crops=image_num_crops,
|
| 1417 |
+
)
|
| 1418 |
+
elif pixel_values_videos is not None:
|
| 1419 |
+
assert input_ids is not None
|
| 1420 |
+
images, token_pooling = self.build_batched_videos(
|
| 1421 |
+
input_ids=input_ids,
|
| 1422 |
+
pixel_values_videos=pixel_values_videos,
|
| 1423 |
+
video_token_pooling=video_token_pooling,
|
| 1424 |
+
video_grids=video_grids,
|
| 1425 |
+
)
|
| 1426 |
+
else:
|
| 1427 |
+
images, token_pooling = None, None
|
| 1428 |
+
return images, token_pooling
|
| 1429 |
+
|
| 1430 |
+
def build_input_embeddings(
|
| 1431 |
+
self,
|
| 1432 |
+
input_ids: torch.LongTensor,
|
| 1433 |
+
images: Optional[torch.FloatTensor] = None, # image inputs
|
| 1434 |
+
token_pooling: Optional[torch.LongTensor] = None,
|
| 1435 |
+
) -> tuple[torch.Tensor, Optional[torch.Tensor]]:
|
| 1436 |
+
|
| 1437 |
+
# Get embeddings of input.
|
| 1438 |
+
# shape: (batch_size, seq_len, d_model)
|
| 1439 |
+
input_ids = input_ids * (input_ids != -1).to(input_ids.dtype)
|
| 1440 |
+
x = self.transformer.wte(input_ids)
|
| 1441 |
+
|
| 1442 |
+
image_features: Optional[torch.FloatTensor] = None
|
| 1443 |
+
if images is not None:
|
| 1444 |
+
image_features = self.vision_backbone(images, token_pooling).to(x.device)
|
| 1445 |
+
is_image_patch = input_ids.view(-1) == self.config.image_patch_id
|
| 1446 |
+
assert is_image_patch.sum() == len(image_features)
|
| 1447 |
+
x.view(-1, x.shape[-1])[is_image_patch] += image_features
|
| 1448 |
+
|
| 1449 |
+
# shape: (batch_size, seq_len, d_model)
|
| 1450 |
+
x = self.transformer.emb_drop(x) # type: ignore
|
| 1451 |
+
|
| 1452 |
+
return x, image_features
|
| 1453 |
+
|
| 1454 |
+
@can_return_tuple
|
| 1455 |
+
def forward(
|
| 1456 |
+
self,
|
| 1457 |
+
input_ids: Optional[torch.LongTensor] = None,
|
| 1458 |
+
pixel_values: Optional[torch.FloatTensor] = None,
|
| 1459 |
+
image_token_pooling: Optional[torch.Tensor] = None,
|
| 1460 |
+
image_grids: Optional[torch.Tensor] = None,
|
| 1461 |
+
image_num_crops: Optional[torch.Tensor] = None,
|
| 1462 |
+
pixel_values_videos: Optional[torch.Tensor] = None,
|
| 1463 |
+
video_token_pooling: Optional[torch.Tensor] = None,
|
| 1464 |
+
video_grids: Optional[torch.Tensor] = None,
|
| 1465 |
+
attention_mask: Optional[torch.Tensor] = None,
|
| 1466 |
+
position_ids: Optional[torch.Tensor] = None,
|
| 1467 |
+
past_key_values: Optional[Cache] = None,
|
| 1468 |
+
token_type_ids: Optional[torch.LongTensor] = None,
|
| 1469 |
+
inputs_embeds: Optional[torch.FloatTensor] = None,
|
| 1470 |
+
use_cache: Optional[bool] = None,
|
| 1471 |
+
output_attentions: Optional[bool] = None,
|
| 1472 |
+
output_hidden_states: Optional[bool] = None,
|
| 1473 |
+
cache_position: Optional[torch.LongTensor] = None,
|
| 1474 |
+
**kwargs: Unpack[TransformersKwargs],
|
| 1475 |
+
) -> Union[tuple, Molmo2ModelOutputWithPast]:
|
| 1476 |
+
|
| 1477 |
+
output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
|
| 1478 |
+
output_hidden_states = (
|
| 1479 |
+
output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
|
| 1480 |
+
)
|
| 1481 |
+
use_cache = use_cache if use_cache is not None else self.config.use_cache
|
| 1482 |
+
|
| 1483 |
+
if (input_ids is None) ^ (inputs_embeds is not None):
|
| 1484 |
+
raise ValueError("You must specify exactly one of input_ids or inputs_embeds")
|
| 1485 |
+
|
| 1486 |
+
images, token_pooling = self.merge_visual_inputs(
|
| 1487 |
+
input_ids=input_ids,
|
| 1488 |
+
pixel_values=pixel_values,
|
| 1489 |
+
image_token_pooling=image_token_pooling,
|
| 1490 |
+
image_grids=image_grids,
|
| 1491 |
+
image_num_crops=image_num_crops,
|
| 1492 |
+
pixel_values_videos=pixel_values_videos,
|
| 1493 |
+
video_token_pooling=video_token_pooling,
|
| 1494 |
+
video_grids=video_grids,
|
| 1495 |
+
)
|
| 1496 |
+
|
| 1497 |
+
if images is not None and inputs_embeds is not None:
|
| 1498 |
+
raise ValueError(
|
| 1499 |
+
"You cannot specify both images and inputs_embeds at the same time."
|
| 1500 |
+
)
|
| 1501 |
+
|
| 1502 |
+
if inputs_embeds is None:
|
| 1503 |
+
inputs_embeds, image_features = self.build_input_embeddings(
|
| 1504 |
+
input_ids, images, token_pooling,
|
| 1505 |
+
)
|
| 1506 |
+
|
| 1507 |
+
if cache_position is None:
|
| 1508 |
+
past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0
|
| 1509 |
+
cache_position = torch.arange(
|
| 1510 |
+
past_seen_tokens,
|
| 1511 |
+
past_seen_tokens + inputs_embeds.shape[1],
|
| 1512 |
+
device=inputs_embeds.device,
|
| 1513 |
+
)
|
| 1514 |
+
|
| 1515 |
+
# Adapted from transformers.models.gemma3.modeling_gemma3
|
| 1516 |
+
# It may already have been prepared by e.g. `generate`
|
| 1517 |
+
if not isinstance(causal_mask_mapping := attention_mask, dict):
|
| 1518 |
+
# Prepare mask arguments
|
| 1519 |
+
mask_kwargs = {
|
| 1520 |
+
"config": self.config.get_text_config(),
|
| 1521 |
+
"input_embeds": inputs_embeds,
|
| 1522 |
+
"attention_mask": attention_mask,
|
| 1523 |
+
"cache_position": cache_position,
|
| 1524 |
+
"past_key_values": past_key_values,
|
| 1525 |
+
"position_ids": position_ids,
|
| 1526 |
+
}
|
| 1527 |
+
|
| 1528 |
+
# NOTE: this `is_prefill` logic is not flawless, it fails when we're using a cache eagerly initialized
|
| 1529 |
+
# (e.g. compiled prefill) AND `images` are not provided. Determining prefill in that case requires
|
| 1530 |
+
# checking data values, which is not compile-compatible.
|
| 1531 |
+
is_prefill = (
|
| 1532 |
+
not use_cache
|
| 1533 |
+
or past_key_values is None
|
| 1534 |
+
or not past_key_values.is_initialized
|
| 1535 |
+
or images is not None
|
| 1536 |
+
)
|
| 1537 |
+
if token_type_ids is not None and is_prefill:
|
| 1538 |
+
# We need to pass an additional mask function to account for token type ids, and it needs to be an `or`
|
| 1539 |
+
mask_kwargs["or_mask_function"] = token_type_ids_mask_function(
|
| 1540 |
+
token_type_ids.to(cache_position.device)
|
| 1541 |
+
)
|
| 1542 |
+
|
| 1543 |
+
# Create the mask
|
| 1544 |
+
causal_mask_mapping = create_causal_mask(**mask_kwargs)
|
| 1545 |
+
|
| 1546 |
+
outputs = self.transformer(
|
| 1547 |
+
attention_mask=causal_mask_mapping,
|
| 1548 |
+
position_ids=position_ids,
|
| 1549 |
+
past_key_values=past_key_values,
|
| 1550 |
+
inputs_embeds=inputs_embeds,
|
| 1551 |
+
use_cache=use_cache,
|
| 1552 |
+
output_attentions=output_attentions,
|
| 1553 |
+
output_hidden_states=output_hidden_states,
|
| 1554 |
+
cache_position=cache_position,
|
| 1555 |
+
**kwargs,
|
| 1556 |
+
)
|
| 1557 |
+
|
| 1558 |
+
return Molmo2ModelOutputWithPast(
|
| 1559 |
+
last_hidden_state=outputs.last_hidden_state,
|
| 1560 |
+
past_key_values=outputs.past_key_values,
|
| 1561 |
+
hidden_states=outputs.hidden_states,
|
| 1562 |
+
attentions=outputs.attentions,
|
| 1563 |
+
image_hidden_states=image_features if images is not None else None,
|
| 1564 |
+
)
|
| 1565 |
+
|
| 1566 |
+
|
| 1567 |
+
class Molmo2ForConditionalGeneration(Molmo2PreTrainedModel, GenerationMixin):
|
| 1568 |
+
_checkpoint_conversion_mapping = {}
|
| 1569 |
+
_tied_weights_keys = [] # Weights are not tied
|
| 1570 |
+
# Reference: fix gemma3 grad acc #37208
|
| 1571 |
+
accepts_loss_kwargs = False
|
| 1572 |
+
config: Molmo2Config
|
| 1573 |
+
|
| 1574 |
+
def __init__(self, config: Molmo2Config):
|
| 1575 |
+
super().__init__(config)
|
| 1576 |
+
|
| 1577 |
+
self.model = Molmo2Model(config)
|
| 1578 |
+
self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
|
| 1579 |
+
self.vocab_size = config.vocab_size
|
| 1580 |
+
|
| 1581 |
+
# Initialize weights and apply final processing
|
| 1582 |
+
self.post_init()
|
| 1583 |
+
|
| 1584 |
+
def get_input_embeddings(self) -> torch.nn.Module:
|
| 1585 |
+
return self.model.transformer.wte
|
| 1586 |
+
|
| 1587 |
+
def set_input_embeddings(self, value: torch.nn.Module) -> None:
|
| 1588 |
+
self.model.transformer.wte = value
|
| 1589 |
+
|
| 1590 |
+
def set_decoder(self, decoder):
|
| 1591 |
+
self.model.set_decoder(decoder)
|
| 1592 |
+
|
| 1593 |
+
def get_decoder(self):
|
| 1594 |
+
return self.model.get_decoder()
|
| 1595 |
+
|
| 1596 |
+
# Make modules available throught conditional class for BC
|
| 1597 |
+
@property
|
| 1598 |
+
def language_model(self) -> torch.nn.Module:
|
| 1599 |
+
return self.model.transformer
|
| 1600 |
+
|
| 1601 |
+
@property
|
| 1602 |
+
def vision_backbone(self) -> torch.nn.Module:
|
| 1603 |
+
return self.model.vision_backbone
|
| 1604 |
+
|
| 1605 |
+
@can_return_tuple
|
| 1606 |
+
def forward(
|
| 1607 |
+
self,
|
| 1608 |
+
input_ids: torch.LongTensor = None,
|
| 1609 |
+
pixel_values: Optional[torch.Tensor] = None,
|
| 1610 |
+
image_token_pooling: Optional[torch.Tensor] = None,
|
| 1611 |
+
image_grids: Optional[torch.Tensor] = None,
|
| 1612 |
+
image_num_crops: Optional[torch.Tensor] = None,
|
| 1613 |
+
pixel_values_videos: Optional[torch.Tensor] = None,
|
| 1614 |
+
video_token_pooling: Optional[torch.Tensor] = None,
|
| 1615 |
+
video_grids: Optional[torch.Tensor] = None,
|
| 1616 |
+
attention_mask: Optional[torch.Tensor] = None,
|
| 1617 |
+
position_ids: Optional[torch.LongTensor] = None,
|
| 1618 |
+
past_key_values: Optional[list[torch.FloatTensor]] = None,
|
| 1619 |
+
token_type_ids: Optional[torch.LongTensor] = None,
|
| 1620 |
+
inputs_embeds: Optional[torch.FloatTensor] = None,
|
| 1621 |
+
labels: Optional[torch.LongTensor] = None,
|
| 1622 |
+
use_cache: Optional[bool] = None,
|
| 1623 |
+
output_attentions: Optional[bool] = None,
|
| 1624 |
+
output_hidden_states: Optional[bool] = None,
|
| 1625 |
+
cache_position: Optional[torch.LongTensor] = None,
|
| 1626 |
+
logits_to_keep: Union[int, torch.Tensor] = 0,
|
| 1627 |
+
**kwargs: Unpack[TransformersKwargs],
|
| 1628 |
+
) -> Union[tuple, Molmo2CausalLMOutputWithPast]:
|
| 1629 |
+
r"""
|
| 1630 |
+
```python
|
| 1631 |
+
>>> from PIL import Image
|
| 1632 |
+
>>> import requests
|
| 1633 |
+
>>> from transformers import AutoProcessor, Molmo2ForConditionalGeneration
|
| 1634 |
+
|
| 1635 |
+
>>> model = Molmo2ForConditionalGeneration.from_pretrained("...")
|
| 1636 |
+
>>> processor = AutoProcessor.from_pretrained("...")
|
| 1637 |
+
|
| 1638 |
+
>>> prompt = "What's the content of the image?"
|
| 1639 |
+
>>> url = "https://www.ilankelman.org/stopsigns/australia.jpg"
|
| 1640 |
+
>>> image = Image.open(requests.get(url, stream=True).raw)
|
| 1641 |
+
|
| 1642 |
+
>>> messages = [{"role": "user", "content": [{"type": "text", "text": prompt}, {"type": "image", "image": image}]}]
|
| 1643 |
+
|
| 1644 |
+
>>> inputs = processor.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt", return_dict=True)
|
| 1645 |
+
|
| 1646 |
+
>>> # Generate
|
| 1647 |
+
>>> generated_ids = model.generate(**inputs, max_new_tokens=15)
|
| 1648 |
+
>>> generated_tokens = generated_ids[:, inputs['input_ids'].size(1):]
|
| 1649 |
+
>>> processor.post_process_image_text_to_text(generated_tokens, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
|
| 1650 |
+
"The image shows a bustling street scene in what appears to be a Chinatown area. There's ..."
|
| 1651 |
+
```"""
|
| 1652 |
+
outputs = self.model(
|
| 1653 |
+
input_ids=input_ids,
|
| 1654 |
+
pixel_values=pixel_values,
|
| 1655 |
+
image_token_pooling=image_token_pooling,
|
| 1656 |
+
image_grids=image_grids,
|
| 1657 |
+
image_num_crops=image_num_crops,
|
| 1658 |
+
pixel_values_videos=pixel_values_videos,
|
| 1659 |
+
video_token_pooling=video_token_pooling,
|
| 1660 |
+
video_grids=video_grids,
|
| 1661 |
+
attention_mask=attention_mask,
|
| 1662 |
+
position_ids=position_ids,
|
| 1663 |
+
past_key_values=past_key_values,
|
| 1664 |
+
token_type_ids=token_type_ids,
|
| 1665 |
+
inputs_embeds=inputs_embeds,
|
| 1666 |
+
use_cache=use_cache,
|
| 1667 |
+
output_attentions=output_attentions,
|
| 1668 |
+
output_hidden_states=output_hidden_states,
|
| 1669 |
+
cache_position=cache_position,
|
| 1670 |
+
**kwargs,
|
| 1671 |
+
)
|
| 1672 |
+
|
| 1673 |
+
hidden_states = outputs.last_hidden_state
|
| 1674 |
+
# Only compute necessary logits, and do not upcast them to float if we are not computing the loss
|
| 1675 |
+
slice_indices = slice(-logits_to_keep, None) if isinstance(logits_to_keep, int) else logits_to_keep
|
| 1676 |
+
logits = self.lm_head(hidden_states[:, slice_indices, :])
|
| 1677 |
+
|
| 1678 |
+
loss = None
|
| 1679 |
+
if labels is not None:
|
| 1680 |
+
loss = self.loss_function(logits=logits, labels=labels, vocab_size=self.vocab_size)
|
| 1681 |
+
|
| 1682 |
+
return Molmo2CausalLMOutputWithPast(
|
| 1683 |
+
loss=loss,
|
| 1684 |
+
logits=logits,
|
| 1685 |
+
past_key_values=outputs.past_key_values,
|
| 1686 |
+
hidden_states=outputs.hidden_states,
|
| 1687 |
+
attentions=outputs.attentions,
|
| 1688 |
+
image_hidden_states=outputs.image_hidden_states,
|
| 1689 |
+
)
|
| 1690 |
+
|
| 1691 |
+
def prepare_inputs_for_generation(
|
| 1692 |
+
self,
|
| 1693 |
+
input_ids: torch.LongTensor,
|
| 1694 |
+
past_key_values: Optional[list[torch.FloatTensor]] = None,
|
| 1695 |
+
inputs_embeds: Optional[torch.FloatTensor] = None,
|
| 1696 |
+
pixel_values: Optional[torch.FloatTensor] = None,
|
| 1697 |
+
image_token_pooling: Optional[torch.Tensor] = None,
|
| 1698 |
+
image_grids: Optional[torch.Tensor] = None,
|
| 1699 |
+
image_num_crops: Optional[torch.Tensor] = None,
|
| 1700 |
+
pixel_values_videos: Optional[torch.Tensor] = None,
|
| 1701 |
+
video_token_pooling: Optional[torch.Tensor] = None,
|
| 1702 |
+
video_grids: Optional[torch.Tensor] = None,
|
| 1703 |
+
attention_mask: Optional[torch.Tensor] = None,
|
| 1704 |
+
token_type_ids: Optional[torch.LongTensor] = None,
|
| 1705 |
+
cache_position: Optional[torch.LongTensor] = None,
|
| 1706 |
+
logits_to_keep: Optional[Union[int, torch.Tensor]] = None,
|
| 1707 |
+
**kwargs,
|
| 1708 |
+
):
|
| 1709 |
+
|
| 1710 |
+
model_inputs = super().prepare_inputs_for_generation(
|
| 1711 |
+
input_ids,
|
| 1712 |
+
past_key_values=past_key_values,
|
| 1713 |
+
inputs_embeds=inputs_embeds,
|
| 1714 |
+
attention_mask=attention_mask,
|
| 1715 |
+
cache_position=cache_position,
|
| 1716 |
+
logits_to_keep=logits_to_keep,
|
| 1717 |
+
token_type_ids=token_type_ids,
|
| 1718 |
+
**kwargs,
|
| 1719 |
+
)
|
| 1720 |
+
|
| 1721 |
+
if cache_position[0] == 0:
|
| 1722 |
+
model_inputs["pixel_values"] = pixel_values
|
| 1723 |
+
model_inputs["image_token_pooling"] = image_token_pooling
|
| 1724 |
+
model_inputs["image_grids"] = image_grids
|
| 1725 |
+
model_inputs["image_num_crops"] = image_num_crops
|
| 1726 |
+
model_inputs["pixel_values_videos"] = pixel_values_videos
|
| 1727 |
+
model_inputs["video_token_pooling"] = video_token_pooling
|
| 1728 |
+
model_inputs["video_grids"] = video_grids
|
| 1729 |
+
|
| 1730 |
+
return model_inputs
|
| 1731 |
+
|
| 1732 |
+
# Adapted from transformers.models.gemma3.modeling_gemma3
|
| 1733 |
+
@staticmethod
|
| 1734 |
+
def create_masks_for_generate(
|
| 1735 |
+
config: PretrainedConfig,
|
| 1736 |
+
input_embeds: torch.Tensor,
|
| 1737 |
+
attention_mask: Optional[torch.Tensor],
|
| 1738 |
+
cache_position: torch.Tensor,
|
| 1739 |
+
past_key_values: Optional[Cache],
|
| 1740 |
+
position_ids: Optional[torch.Tensor],
|
| 1741 |
+
token_type_ids: Optional[torch.Tensor] = None,
|
| 1742 |
+
**kwargs,
|
| 1743 |
+
) -> dict:
|
| 1744 |
+
# Prepare mask arguments
|
| 1745 |
+
mask_kwargs = {
|
| 1746 |
+
"config": config.get_text_config(),
|
| 1747 |
+
"input_embeds": input_embeds,
|
| 1748 |
+
"attention_mask": attention_mask,
|
| 1749 |
+
"cache_position": cache_position,
|
| 1750 |
+
"past_key_values": past_key_values,
|
| 1751 |
+
"position_ids": position_ids,
|
| 1752 |
+
}
|
| 1753 |
+
# Add the token type ids mask for generate as well
|
| 1754 |
+
if token_type_ids is not None and input_embeds.shape[1] != 1:
|
| 1755 |
+
# We need to pass an additional mask function to account for token type ids, and it needs to be an `or`
|
| 1756 |
+
mask_kwargs["or_mask_function"] = token_type_ids_mask_function(
|
| 1757 |
+
token_type_ids.to(cache_position.device)
|
| 1758 |
+
)
|
| 1759 |
+
|
| 1760 |
+
return create_masks_for_generate(**mask_kwargs)
|
| 1761 |
+
|
| 1762 |
+
|
| 1763 |
+
# Always register for multi-modal features
|
| 1764 |
+
AutoModelForImageTextToText.register(Molmo2Config, Molmo2ForConditionalGeneration)
|
molmoweb_logo.png
ADDED
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"auto_map": {
|
| 3 |
+
"AutoImageProcessor": "image_processing_molmo2.Molmo2ImageProcessor",
|
| 4 |
+
"AutoProcessor": "processing_molmo2.Molmo2Processor"
|
| 5 |
+
},
|
| 6 |
+
"do_convert_rgb": true,
|
| 7 |
+
"image_mean": [
|
| 8 |
+
0.5,
|
| 9 |
+
0.5,
|
| 10 |
+
0.5
|
| 11 |
+
],
|
| 12 |
+
"image_processor_type": "Molmo2ImageProcessor",
|
| 13 |
+
"image_std": [
|
| 14 |
+
0.5,
|
| 15 |
+
0.5,
|
| 16 |
+
0.5
|
| 17 |
+
],
|
| 18 |
+
"max_crops": 8,
|
| 19 |
+
"overlap_margins": [
|
| 20 |
+
4,
|
| 21 |
+
4
|
| 22 |
+
],
|
| 23 |
+
"patch_size": 14,
|
| 24 |
+
"pooling_size": [
|
| 25 |
+
2,
|
| 26 |
+
2
|
| 27 |
+
],
|
| 28 |
+
"processor_class": "Molmo2Processor",
|
| 29 |
+
"resample": 2,
|
| 30 |
+
"size": {
|
| 31 |
+
"height": 378,
|
| 32 |
+
"width": 378
|
| 33 |
+
}
|
| 34 |
+
}
|
processing_molmo2.py
ADDED
|
@@ -0,0 +1,403 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Processor class for Molmo2.
|
| 3 |
+
"""
|
| 4 |
+
from typing import Optional, Union
|
| 5 |
+
import dataclasses
|
| 6 |
+
|
| 7 |
+
import numpy as np
|
| 8 |
+
|
| 9 |
+
from transformers.image_utils import ImageInput
|
| 10 |
+
from transformers.video_utils import VideoInput
|
| 11 |
+
from transformers.processing_utils import (
|
| 12 |
+
Unpack,
|
| 13 |
+
ProcessingKwargs,
|
| 14 |
+
ProcessorMixin,
|
| 15 |
+
)
|
| 16 |
+
from transformers.feature_extraction_utils import BatchFeature
|
| 17 |
+
from transformers.tokenization_utils_base import TextInput, PreTokenizedInput
|
| 18 |
+
from transformers.utils import logging
|
| 19 |
+
|
| 20 |
+
from transformers import AutoTokenizer
|
| 21 |
+
from .image_processing_molmo2 import Molmo2ImagesKwargs, Molmo2ImageProcessor
|
| 22 |
+
from .video_processing_molmo2 import Molmo2VideoProcessorKwargs, Molmo2VideoProcessor
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
logger = logging.get_logger(__name__)
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
# Special tokens, these should be present in any tokenizer we use since the preprocessor uses them
|
| 29 |
+
IMAGE_PATCH_TOKEN = f"<im_patch>" # Where to insert high-res tokens
|
| 30 |
+
IMAGE_LOW_RES_TOKEN = f"<im_low>" # Where to insert low-res tokens
|
| 31 |
+
IM_START_TOKEN = f"<im_start>"
|
| 32 |
+
LOW_RES_IMAGE_START_TOKEN = f"<low_res_im_start>"
|
| 33 |
+
FRAME_START_TOKEN = f"<frame_start>"
|
| 34 |
+
IM_END_TOKEN = f"<im_end>"
|
| 35 |
+
FRAME_END_TOKEN= f"<frame_end>"
|
| 36 |
+
IM_COL_TOKEN = f"<im_col>"
|
| 37 |
+
IMAGE_PROMPT = "<|image|>"
|
| 38 |
+
VIDEO_PROMPT = "<|video|>"
|
| 39 |
+
|
| 40 |
+
IMAGE_TOKENS = [
|
| 41 |
+
IMAGE_PATCH_TOKEN,
|
| 42 |
+
IM_COL_TOKEN,
|
| 43 |
+
IM_START_TOKEN,
|
| 44 |
+
LOW_RES_IMAGE_START_TOKEN,
|
| 45 |
+
FRAME_START_TOKEN,
|
| 46 |
+
IM_END_TOKEN,
|
| 47 |
+
FRAME_END_TOKEN,
|
| 48 |
+
IMAGE_LOW_RES_TOKEN,
|
| 49 |
+
]
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
class Molmo2ProcessorKwargs(ProcessingKwargs, total=False):
|
| 53 |
+
"""Molmo2 processor kwargs"""
|
| 54 |
+
images_kwargs: Molmo2ImagesKwargs
|
| 55 |
+
videos_kwargs: Molmo2VideoProcessorKwargs
|
| 56 |
+
_defaults = {
|
| 57 |
+
"text_kwargs": {
|
| 58 |
+
"padding": False,
|
| 59 |
+
"return_mm_token_type_ids": True,
|
| 60 |
+
},
|
| 61 |
+
"videos_kwargs": {"return_metadata": True},
|
| 62 |
+
}
|
| 63 |
+
|
| 64 |
+
|
| 65 |
+
class Molmo2Processor(ProcessorMixin):
|
| 66 |
+
attributes = ["image_processor", "video_processor", "tokenizer"]
|
| 67 |
+
optional_attributes = [
|
| 68 |
+
"chat_template",
|
| 69 |
+
"time_mode",
|
| 70 |
+
"image_use_col_tokens",
|
| 71 |
+
"use_single_crop_col_tokens",
|
| 72 |
+
"use_single_crop_start_token",
|
| 73 |
+
"video_use_col_tokens",
|
| 74 |
+
"use_frame_special_tokens",
|
| 75 |
+
]
|
| 76 |
+
image_processor_class = "AutoImageProcessor"
|
| 77 |
+
video_processor_class = "AutoVideoProcessor"
|
| 78 |
+
tokenizer_class = "AutoTokenizer"
|
| 79 |
+
|
| 80 |
+
def __init__(
|
| 81 |
+
self,
|
| 82 |
+
image_processor: Molmo2ImageProcessor = None,
|
| 83 |
+
video_processor: Molmo2VideoProcessor = None,
|
| 84 |
+
tokenizer: AutoTokenizer = None,
|
| 85 |
+
chat_template: Optional[str] = None,
|
| 86 |
+
image_use_col_tokens: Optional[bool] = True,
|
| 87 |
+
use_single_crop_col_tokens: Optional[bool] = None,
|
| 88 |
+
use_single_crop_start_token: Optional[bool] = True,
|
| 89 |
+
video_use_col_tokens: Optional[bool] = False,
|
| 90 |
+
use_frame_special_tokens: Optional[bool] = True,
|
| 91 |
+
**kwargs
|
| 92 |
+
) -> None:
|
| 93 |
+
super().__init__(
|
| 94 |
+
image_processor,
|
| 95 |
+
video_processor,
|
| 96 |
+
tokenizer,
|
| 97 |
+
chat_template=chat_template,
|
| 98 |
+
image_use_col_tokens=image_use_col_tokens,
|
| 99 |
+
use_single_crop_col_tokens=use_single_crop_col_tokens,
|
| 100 |
+
use_single_crop_start_token=use_single_crop_start_token,
|
| 101 |
+
video_use_col_tokens=video_use_col_tokens,
|
| 102 |
+
use_frame_special_tokens=use_frame_special_tokens,
|
| 103 |
+
)
|
| 104 |
+
|
| 105 |
+
self.image_placeholder_token = IMAGE_PROMPT
|
| 106 |
+
self.video_placeholder_token = VIDEO_PROMPT
|
| 107 |
+
self.image_token_ids = [
|
| 108 |
+
tokenizer.convert_tokens_to_ids(token)
|
| 109 |
+
for token in IMAGE_TOKENS
|
| 110 |
+
]
|
| 111 |
+
|
| 112 |
+
def get_image_tokens(self, image_grid: np.ndarray):
|
| 113 |
+
resized_h, resized_w, height, width = image_grid
|
| 114 |
+
per_row = np.full(width, IMAGE_PATCH_TOKEN)
|
| 115 |
+
if self.image_use_col_tokens:
|
| 116 |
+
per_row = np.concatenate([per_row, [IM_COL_TOKEN]], 0)
|
| 117 |
+
joint = [
|
| 118 |
+
[IM_START_TOKEN],
|
| 119 |
+
np.tile(per_row, [height]),
|
| 120 |
+
[IM_END_TOKEN],
|
| 121 |
+
]
|
| 122 |
+
per_row = np.full(resized_w, IMAGE_PATCH_TOKEN)
|
| 123 |
+
use_single_crop_col_tokens = (
|
| 124 |
+
self.image_use_col_tokens
|
| 125 |
+
if self.use_single_crop_col_tokens is None
|
| 126 |
+
else self.use_single_crop_col_tokens
|
| 127 |
+
)
|
| 128 |
+
image_start_token = (
|
| 129 |
+
LOW_RES_IMAGE_START_TOKEN
|
| 130 |
+
if self.use_single_crop_start_token
|
| 131 |
+
else IM_START_TOKEN
|
| 132 |
+
)
|
| 133 |
+
if use_single_crop_col_tokens:
|
| 134 |
+
per_row = np.concatenate([per_row, [IM_COL_TOKEN]], 0)
|
| 135 |
+
joint = [
|
| 136 |
+
[image_start_token],
|
| 137 |
+
np.tile(per_row, [resized_h]),
|
| 138 |
+
[IM_END_TOKEN],
|
| 139 |
+
] + joint
|
| 140 |
+
|
| 141 |
+
return np.concatenate(joint)
|
| 142 |
+
|
| 143 |
+
def get_video_string(
|
| 144 |
+
self,
|
| 145 |
+
video_grid: np.ndarray,
|
| 146 |
+
timestamps: np.ndarray,
|
| 147 |
+
):
|
| 148 |
+
if self.use_frame_special_tokens:
|
| 149 |
+
start_token_id = FRAME_START_TOKEN
|
| 150 |
+
end_token_id = FRAME_END_TOKEN
|
| 151 |
+
else:
|
| 152 |
+
start_token_id = IM_START_TOKEN
|
| 153 |
+
end_token_id = IM_END_TOKEN
|
| 154 |
+
|
| 155 |
+
num_frames, h, w = video_grid
|
| 156 |
+
video_string: str = ""
|
| 157 |
+
for frame_idx, frame_time in enumerate(timestamps):
|
| 158 |
+
# `per-frame-compact` time mode
|
| 159 |
+
prev_space = " " if frame_idx > 0 else ""
|
| 160 |
+
frame_prefix = prev_space + f"{frame_time:.1f} " # explicit whitespace before/after image tokens
|
| 161 |
+
|
| 162 |
+
video_string += frame_prefix
|
| 163 |
+
per_row = np.full(w, IMAGE_PATCH_TOKEN)
|
| 164 |
+
if self.video_use_col_tokens:
|
| 165 |
+
per_row = np.concatenate([per_row, [IM_COL_TOKEN]], 0)
|
| 166 |
+
extra_tokens = np.tile(per_row, [h])
|
| 167 |
+
video_tokens = [
|
| 168 |
+
[start_token_id],
|
| 169 |
+
extra_tokens,
|
| 170 |
+
[end_token_id],
|
| 171 |
+
]
|
| 172 |
+
video_string += "".join(np.concatenate(video_tokens, 0))
|
| 173 |
+
|
| 174 |
+
return video_string
|
| 175 |
+
|
| 176 |
+
def insert_bos(
|
| 177 |
+
self,
|
| 178 |
+
input_ids: np.ndarray,
|
| 179 |
+
attention_mask: np.ndarray,
|
| 180 |
+
bos_token_id: int,
|
| 181 |
+
pad_token_id: int,
|
| 182 |
+
):
|
| 183 |
+
"""
|
| 184 |
+
Args:
|
| 185 |
+
input_ids: [B, S] array with left padding
|
| 186 |
+
attention_mask: [B, S] array (0 for pad, 1 for valid)
|
| 187 |
+
bos_token_id: int
|
| 188 |
+
pad_token_id: int
|
| 189 |
+
Returns:
|
| 190 |
+
input_ids_out: [B, S] or [B, S+1] array with bos inserted if needed
|
| 191 |
+
attention_mask_out: same shape as input_ids_out
|
| 192 |
+
"""
|
| 193 |
+
|
| 194 |
+
need_to_expand = len(input_ids.shape) == 1
|
| 195 |
+
if need_to_expand:
|
| 196 |
+
input_ids = input_ids[None, :]
|
| 197 |
+
attention_mask = attention_mask[None, :]
|
| 198 |
+
|
| 199 |
+
B, S = input_ids.shape
|
| 200 |
+
|
| 201 |
+
# Handle zero-length sequence
|
| 202 |
+
if S == 0:
|
| 203 |
+
new_input_ids = np.full((B, 1), bos_token_id, dtype=input_ids.dtype)
|
| 204 |
+
new_attention_mask = np.ones((B, 1), dtype=attention_mask.dtype)
|
| 205 |
+
if need_to_expand:
|
| 206 |
+
new_input_ids = new_input_ids[0]
|
| 207 |
+
new_attention_mask = new_attention_mask[0]
|
| 208 |
+
return new_input_ids, new_attention_mask
|
| 209 |
+
|
| 210 |
+
first_valid_index = (attention_mask == 1).argmax(axis=-1) # [B]
|
| 211 |
+
bos_already_present = np.all(input_ids[np.arange(B), first_valid_index] == bos_token_id)
|
| 212 |
+
|
| 213 |
+
if bos_already_present:
|
| 214 |
+
if need_to_expand:
|
| 215 |
+
input_ids = input_ids[0]
|
| 216 |
+
attention_mask = attention_mask[0]
|
| 217 |
+
return input_ids, attention_mask
|
| 218 |
+
else:
|
| 219 |
+
new_input_ids = np.full((B, S+1), pad_token_id, dtype=input_ids.dtype)
|
| 220 |
+
new_attention_mask = np.zeros((B, S+1), dtype=attention_mask.dtype)
|
| 221 |
+
|
| 222 |
+
src_idx = np.tile(np.arange(S), (B, 1)) # [B, S]
|
| 223 |
+
valid_mask = src_idx >= first_valid_index[:, None] # [B, S]
|
| 224 |
+
tgt_idx = src_idx + 1 # shit right
|
| 225 |
+
batch_idx = np.tile(np.arange(B)[:, None], (1, S)) # [B, S]
|
| 226 |
+
|
| 227 |
+
# flatten valid_positions
|
| 228 |
+
flat_vals = input_ids[valid_mask]
|
| 229 |
+
flat_batch = batch_idx[valid_mask]
|
| 230 |
+
flat_tgt = tgt_idx[valid_mask]
|
| 231 |
+
|
| 232 |
+
new_input_ids[flat_batch, flat_tgt] = flat_vals
|
| 233 |
+
new_attention_mask[flat_batch, flat_tgt] = 1
|
| 234 |
+
|
| 235 |
+
insert_pos = first_valid_index
|
| 236 |
+
new_input_ids[np.arange(B), insert_pos] = bos_token_id
|
| 237 |
+
new_attention_mask[np.arange(B), insert_pos] = 1
|
| 238 |
+
|
| 239 |
+
if need_to_expand:
|
| 240 |
+
new_input_ids = new_input_ids[0]
|
| 241 |
+
new_attention_mask = new_attention_mask[0]
|
| 242 |
+
|
| 243 |
+
return new_input_ids, new_attention_mask
|
| 244 |
+
|
| 245 |
+
def __call__(
|
| 246 |
+
self,
|
| 247 |
+
text: Union[TextInput, PreTokenizedInput, list[TextInput], list[PreTokenizedInput]] = None,
|
| 248 |
+
images: ImageInput = None,
|
| 249 |
+
videos: VideoInput = None,
|
| 250 |
+
**kwargs: Unpack[Molmo2ProcessorKwargs],
|
| 251 |
+
) -> BatchFeature:
|
| 252 |
+
"""
|
| 253 |
+
|
| 254 |
+
Args:
|
| 255 |
+
text (`str`, `list[str]`, `list[list[str]]`):
|
| 256 |
+
The sequence or batch of sequences to be encoded. Each sequence can be a string or a list of strings
|
| 257 |
+
(pretokenized string). If the sequences are provided as list of strings (pretokenized), you must set
|
| 258 |
+
`is_split_into_words=True` (to lift the ambiguity with a batch of sequences).
|
| 259 |
+
images (`PIL.Image.Image`, `np.ndarray`, `torch.Tensor`, `list[PIL.Image.Image]`, `list[np.ndarray]`, `list[torch.Tensor]`):
|
| 260 |
+
The image or batch of images to be prepared. Each image can be a PIL image, NumPy array or PyTorch
|
| 261 |
+
tensor. Both channels-first and channels-last formats are supported.
|
| 262 |
+
videos (`dict[str, Any]` or `list[dict[str, Any]]`):
|
| 263 |
+
The video or batch of videos to be prepared. Each video can be a dictionary with the following keys:
|
| 264 |
+
- `"frames"`: `np.ndarray` of shape (T, H, W, 3)
|
| 265 |
+
- `"timestamps"`: `np.ndarray` of shape (T,)
|
| 266 |
+
- `"sampled_fps"`: `float` (optional)
|
| 267 |
+
- `"sampling_augmentation"`: `str` (optional)
|
| 268 |
+
return_tensors (`str` or [`~utils.TensorType`], *optional*):
|
| 269 |
+
If set, will return tensors of a particular framework. Acceptable values are:
|
| 270 |
+
- `'tf'`: Return TensorFlow `tf.constant` objects.
|
| 271 |
+
- `'pt'`: Return PyTorch `torch.Tensor` objects.
|
| 272 |
+
- `'np'`: Return NumPy `np.ndarray` objects.
|
| 273 |
+
- `'jax'`: Return JAX `jnp.ndarray` objects.
|
| 274 |
+
|
| 275 |
+
Returns:
|
| 276 |
+
`BatchFeature`: A [`BatchFeature`] with the following fields:
|
| 277 |
+
- **input_ids** -- List of token ids to be fed to a model. Returned when `text` is not `None`.
|
| 278 |
+
- **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when
|
| 279 |
+
`return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names` and if `text` is not `None`).
|
| 280 |
+
- **pixel_values** -- Pixel values to be fed to a model. Returned when `images` is not `None`.
|
| 281 |
+
- **image_token_pooling** -- Indices of the patches in `image_grids` to pool for each token in `image_tokens`.
|
| 282 |
+
Returned when `images` is not `None`.
|
| 283 |
+
- **image_grids** -- Grids of images. Returned when `images` is not `None`.
|
| 284 |
+
- **image_num_crops** -- Number of crops for each image. Returned when `images` is not `None`.
|
| 285 |
+
- **pixel_values_videos** -- Pixel values of videos to be fed to a model. Returned when `videos` is not `None`.
|
| 286 |
+
- **video_token_pooling** -- Indices of the patches in `video_grids` to pool for each token in `video_tokens`.
|
| 287 |
+
Returned when `videos` is not `None`.
|
| 288 |
+
- **video_grids** -- Grids of videos. Returned when `videos` is not `None`.
|
| 289 |
+
"""
|
| 290 |
+
|
| 291 |
+
output_kwargs = self._merge_kwargs(
|
| 292 |
+
Molmo2ProcessorKwargs,
|
| 293 |
+
tokenizer_init_kwargs=self.tokenizer.init_kwargs,
|
| 294 |
+
**kwargs,
|
| 295 |
+
)
|
| 296 |
+
|
| 297 |
+
if images is not None:
|
| 298 |
+
image_inputs = self.image_processor(images, **output_kwargs["images_kwargs"])
|
| 299 |
+
image_grids = image_inputs["image_grids"]
|
| 300 |
+
else:
|
| 301 |
+
image_inputs = {}
|
| 302 |
+
image_grids = None
|
| 303 |
+
|
| 304 |
+
if videos is not None:
|
| 305 |
+
videos_inputs = self.video_processor(videos=videos, **output_kwargs["videos_kwargs"])
|
| 306 |
+
video_grids = videos_inputs["video_grids"]
|
| 307 |
+
# If user has not requested video metadata, pop it
|
| 308 |
+
if "return_metadata" not in kwargs:
|
| 309 |
+
video_metadata = videos_inputs.pop("video_metadata")
|
| 310 |
+
else:
|
| 311 |
+
video_metadata = videos_inputs["video_metadata"]
|
| 312 |
+
else:
|
| 313 |
+
videos_inputs = {}
|
| 314 |
+
video_grids = None
|
| 315 |
+
|
| 316 |
+
if not isinstance(text, list):
|
| 317 |
+
text = [text]
|
| 318 |
+
|
| 319 |
+
text = text.copy() # below lines change text in-place
|
| 320 |
+
|
| 321 |
+
if image_grids is not None:
|
| 322 |
+
index = 0
|
| 323 |
+
for i in range(len(text)):
|
| 324 |
+
num_images = text[i].count(self.image_placeholder_token)
|
| 325 |
+
image_grids_i = image_grids[index:index+num_images]
|
| 326 |
+
for image_grid in image_grids_i:
|
| 327 |
+
image_tokens = self.get_image_tokens(image_grid)
|
| 328 |
+
image_string = "".join(image_tokens)
|
| 329 |
+
text[i] = text[i].replace(self.image_placeholder_token, image_string, 1)
|
| 330 |
+
index += num_images
|
| 331 |
+
|
| 332 |
+
if video_grids is not None:
|
| 333 |
+
index = 0
|
| 334 |
+
for i in range(len(text)):
|
| 335 |
+
num_videos = text[i].count(self.video_placeholder_token)
|
| 336 |
+
assert num_videos in {0, 1}, "At most one video is supported for now"
|
| 337 |
+
video_grids_i = video_grids[index:index+num_videos]
|
| 338 |
+
metadata_i = video_metadata[index:index+num_videos]
|
| 339 |
+
for video_grid, metadata in zip(video_grids_i, metadata_i):
|
| 340 |
+
video_string = self.get_video_string(
|
| 341 |
+
video_grid,
|
| 342 |
+
metadata.timestamps,
|
| 343 |
+
)
|
| 344 |
+
text[i] = text[i].replace(self.video_placeholder_token, video_string, 1)
|
| 345 |
+
index += num_videos
|
| 346 |
+
|
| 347 |
+
return_tensors = output_kwargs["text_kwargs"].pop("return_tensors", None)
|
| 348 |
+
return_mm_token_type_ids = output_kwargs["text_kwargs"].pop("return_mm_token_type_ids", False)
|
| 349 |
+
text_inputs = self.tokenizer(text, **output_kwargs["text_kwargs"])
|
| 350 |
+
|
| 351 |
+
input_ids = text_inputs["input_ids"]
|
| 352 |
+
attention_mask = text_inputs["attention_mask"]
|
| 353 |
+
|
| 354 |
+
input_ids = np.array(input_ids)
|
| 355 |
+
attention_mask = np.array(attention_mask)
|
| 356 |
+
|
| 357 |
+
bos = self.tokenizer.bos_token_id or self.tokenizer.eos_token_id
|
| 358 |
+
input_ids, attention_mask = self.insert_bos(
|
| 359 |
+
input_ids, attention_mask, bos, self.tokenizer.pad_token_id
|
| 360 |
+
)
|
| 361 |
+
|
| 362 |
+
if return_mm_token_type_ids:
|
| 363 |
+
image_tokens = np.array(self.image_token_ids).astype(input_ids.dtype)
|
| 364 |
+
token_type_ids = np.any(input_ids[:, :, None] == image_tokens[None, None, :], axis=-1)
|
| 365 |
+
text_inputs["token_type_ids"] = token_type_ids.tolist()
|
| 366 |
+
|
| 367 |
+
text_inputs["input_ids"] = input_ids.tolist()
|
| 368 |
+
text_inputs["attention_mask"] = attention_mask.tolist()
|
| 369 |
+
|
| 370 |
+
return BatchFeature(
|
| 371 |
+
data={**text_inputs, **image_inputs, **videos_inputs},
|
| 372 |
+
tensor_type=return_tensors,
|
| 373 |
+
)
|
| 374 |
+
|
| 375 |
+
def post_process_image_text_to_text(
|
| 376 |
+
self, generated_outputs, skip_special_tokens=True, clean_up_tokenization_spaces=False, **kwargs
|
| 377 |
+
):
|
| 378 |
+
"""
|
| 379 |
+
Post-process the output of the model to decode the text.
|
| 380 |
+
|
| 381 |
+
Args:
|
| 382 |
+
generated_outputs (`torch.Tensor` or `np.ndarray`):
|
| 383 |
+
The output of the model `generate` function. The output is expected to be a tensor of shape `(batch_size, sequence_length)`
|
| 384 |
+
or `(sequence_length,)`.
|
| 385 |
+
skip_special_tokens (`bool`, *optional*, defaults to `True`):
|
| 386 |
+
Whether or not to remove special tokens in the output. Argument passed to the tokenizer's `batch_decode` method.
|
| 387 |
+
clean_up_tokenization_spaces (`bool`, *optional*, defaults to `False`):
|
| 388 |
+
Whether or not to clean up the tokenization spaces. Argument passed to the tokenizer's `batch_decode` method.
|
| 389 |
+
**kwargs:
|
| 390 |
+
Additional arguments to be passed to the tokenizer's `batch_decode method`.
|
| 391 |
+
|
| 392 |
+
Returns:
|
| 393 |
+
`list[str]`: The decoded text.
|
| 394 |
+
"""
|
| 395 |
+
return self.tokenizer.batch_decode(
|
| 396 |
+
generated_outputs,
|
| 397 |
+
skip_special_tokens=skip_special_tokens,
|
| 398 |
+
clean_up_tokenization_spaces=clean_up_tokenization_spaces,
|
| 399 |
+
**kwargs,
|
| 400 |
+
)
|
| 401 |
+
|
| 402 |
+
|
| 403 |
+
Molmo2Processor.register_for_auto_class()
|
processor_config.json
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"auto_map": {
|
| 3 |
+
"AutoProcessor": "processing_molmo2.Molmo2Processor"
|
| 4 |
+
},
|
| 5 |
+
"image_use_col_tokens": true,
|
| 6 |
+
"processor_class": "Molmo2Processor",
|
| 7 |
+
"use_frame_special_tokens": false,
|
| 8 |
+
"use_single_crop_col_tokens": null,
|
| 9 |
+
"use_single_crop_start_token": false,
|
| 10 |
+
"video_use_col_tokens": false
|
| 11 |
+
}
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,296 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"additional_special_tokens": [
|
| 3 |
+
"|<EXTRA_TOKENS_0>|",
|
| 4 |
+
"|<EXTRA_TOKENS_1>|",
|
| 5 |
+
"|<EXTRA_TOKENS_2>|",
|
| 6 |
+
"|<EXTRA_TOKENS_3>|",
|
| 7 |
+
"|<EXTRA_TOKENS_4>|",
|
| 8 |
+
"|<EXTRA_TOKENS_5>|",
|
| 9 |
+
"|<EXTRA_TOKENS_6>|",
|
| 10 |
+
"|<EXTRA_TOKENS_7>|",
|
| 11 |
+
"|<EXTRA_TOKENS_8>|",
|
| 12 |
+
"|<EXTRA_TOKENS_9>|",
|
| 13 |
+
"|<EXTRA_TOKENS_10>|",
|
| 14 |
+
"|<EXTRA_TOKENS_11>|",
|
| 15 |
+
"|<EXTRA_TOKENS_12>|",
|
| 16 |
+
"|<EXTRA_TOKENS_13>|",
|
| 17 |
+
"|<EXTRA_TOKENS_14>|",
|
| 18 |
+
"|<EXTRA_TOKENS_15>|",
|
| 19 |
+
"|<EXTRA_TOKENS_16>|",
|
| 20 |
+
"|<EXTRA_TOKENS_17>|",
|
| 21 |
+
"|<EXTRA_TOKENS_18>|",
|
| 22 |
+
"|<EXTRA_TOKENS_19>|",
|
| 23 |
+
"|<EXTRA_TOKENS_20>|",
|
| 24 |
+
"|<EXTRA_TOKENS_21>|",
|
| 25 |
+
"|<EXTRA_TOKENS_22>|",
|
| 26 |
+
"|<EXTRA_TOKENS_23>|",
|
| 27 |
+
"|<EXTRA_TOKENS_24>|",
|
| 28 |
+
"|<EXTRA_TOKENS_25>|",
|
| 29 |
+
"|<EXTRA_TOKENS_26>|",
|
| 30 |
+
"|<EXTRA_TOKENS_27>|",
|
| 31 |
+
"|<EXTRA_TOKENS_28>|",
|
| 32 |
+
"|<EXTRA_TOKENS_29>|",
|
| 33 |
+
"|<EXTRA_TOKENS_30>|",
|
| 34 |
+
"|<EXTRA_TOKENS_31>|",
|
| 35 |
+
"|<EXTRA_TOKENS_32>|",
|
| 36 |
+
"|<EXTRA_TOKENS_33>|",
|
| 37 |
+
"|<EXTRA_TOKENS_34>|",
|
| 38 |
+
"|<EXTRA_TOKENS_35>|",
|
| 39 |
+
"|<EXTRA_TOKENS_36>|",
|
| 40 |
+
"|<EXTRA_TOKENS_37>|",
|
| 41 |
+
"|<EXTRA_TOKENS_38>|",
|
| 42 |
+
"|<EXTRA_TOKENS_39>|",
|
| 43 |
+
"|<EXTRA_TOKENS_40>|",
|
| 44 |
+
"|<EXTRA_TOKENS_41>|",
|
| 45 |
+
"|<EXTRA_TOKENS_42>|",
|
| 46 |
+
"|<EXTRA_TOKENS_43>|",
|
| 47 |
+
"|<EXTRA_TOKENS_44>|",
|
| 48 |
+
"|<EXTRA_TOKENS_45>|",
|
| 49 |
+
"|<EXTRA_TOKENS_46>|",
|
| 50 |
+
"|<EXTRA_TOKENS_47>|",
|
| 51 |
+
"|<EXTRA_TOKENS_48>|",
|
| 52 |
+
"|<EXTRA_TOKENS_49>|",
|
| 53 |
+
"|<EXTRA_TOKENS_50>|",
|
| 54 |
+
"|<EXTRA_TOKENS_51>|",
|
| 55 |
+
"|<EXTRA_TOKENS_52>|",
|
| 56 |
+
"|<EXTRA_TOKENS_53>|",
|
| 57 |
+
"|<EXTRA_TOKENS_54>|",
|
| 58 |
+
"|<EXTRA_TOKENS_55>|",
|
| 59 |
+
"|<EXTRA_TOKENS_56>|",
|
| 60 |
+
"|<EXTRA_TOKENS_57>|",
|
| 61 |
+
"|<EXTRA_TOKENS_58>|",
|
| 62 |
+
"|<EXTRA_TOKENS_59>|",
|
| 63 |
+
"|<EXTRA_TOKENS_60>|",
|
| 64 |
+
"|<EXTRA_TOKENS_61>|",
|
| 65 |
+
"|<EXTRA_TOKENS_62>|",
|
| 66 |
+
"|<EXTRA_TOKENS_63>|",
|
| 67 |
+
"|<EXTRA_TOKENS_64>|",
|
| 68 |
+
"|<EXTRA_TOKENS_65>|",
|
| 69 |
+
"|<EXTRA_TOKENS_66>|",
|
| 70 |
+
"|<EXTRA_TOKENS_67>|",
|
| 71 |
+
"|<EXTRA_TOKENS_68>|",
|
| 72 |
+
"|<EXTRA_TOKENS_69>|",
|
| 73 |
+
"|<EXTRA_TOKENS_70>|",
|
| 74 |
+
"|<EXTRA_TOKENS_71>|",
|
| 75 |
+
"|<EXTRA_TOKENS_72>|",
|
| 76 |
+
"|<EXTRA_TOKENS_73>|",
|
| 77 |
+
"|<EXTRA_TOKENS_74>|",
|
| 78 |
+
"|<EXTRA_TOKENS_75>|",
|
| 79 |
+
"|<EXTRA_TOKENS_76>|",
|
| 80 |
+
"|<EXTRA_TOKENS_77>|",
|
| 81 |
+
"|<EXTRA_TOKENS_78>|",
|
| 82 |
+
"|<EXTRA_TOKENS_79>|",
|
| 83 |
+
"|<EXTRA_TOKENS_80>|",
|
| 84 |
+
"|<EXTRA_TOKENS_81>|",
|
| 85 |
+
"|<EXTRA_TOKENS_82>|",
|
| 86 |
+
"|<EXTRA_TOKENS_83>|",
|
| 87 |
+
"|<EXTRA_TOKENS_84>|",
|
| 88 |
+
"|<EXTRA_TOKENS_85>|",
|
| 89 |
+
"|<EXTRA_TOKENS_86>|",
|
| 90 |
+
"|<EXTRA_TOKENS_87>|",
|
| 91 |
+
"|<EXTRA_TOKENS_88>|",
|
| 92 |
+
"|<EXTRA_TOKENS_89>|",
|
| 93 |
+
"|<EXTRA_TOKENS_90>|",
|
| 94 |
+
"|<EXTRA_TOKENS_91>|",
|
| 95 |
+
"|<EXTRA_TOKENS_92>|",
|
| 96 |
+
"|<EXTRA_TOKENS_93>|",
|
| 97 |
+
"|<EXTRA_TOKENS_94>|",
|
| 98 |
+
"|<EXTRA_TOKENS_95>|",
|
| 99 |
+
"|<EXTRA_TOKENS_96>|",
|
| 100 |
+
"|<EXTRA_TOKENS_97>|",
|
| 101 |
+
"|<EXTRA_TOKENS_98>|",
|
| 102 |
+
"|<EXTRA_TOKENS_99>|",
|
| 103 |
+
"|<EXTRA_TOKENS_100>|",
|
| 104 |
+
"|<EXTRA_TOKENS_101>|",
|
| 105 |
+
"|<EXTRA_TOKENS_102>|",
|
| 106 |
+
"|<EXTRA_TOKENS_103>|",
|
| 107 |
+
"|<EXTRA_TOKENS_104>|",
|
| 108 |
+
"|<EXTRA_TOKENS_105>|",
|
| 109 |
+
"|<EXTRA_TOKENS_106>|",
|
| 110 |
+
"|<EXTRA_TOKENS_107>|",
|
| 111 |
+
"|<EXTRA_TOKENS_108>|",
|
| 112 |
+
"|<EXTRA_TOKENS_109>|",
|
| 113 |
+
"|<EXTRA_TOKENS_110>|",
|
| 114 |
+
"|<EXTRA_TOKENS_111>|",
|
| 115 |
+
"|<EXTRA_TOKENS_112>|",
|
| 116 |
+
"|<EXTRA_TOKENS_113>|",
|
| 117 |
+
"|<EXTRA_TOKENS_114>|",
|
| 118 |
+
"|<EXTRA_TOKENS_115>|",
|
| 119 |
+
"|<EXTRA_TOKENS_116>|",
|
| 120 |
+
"|<EXTRA_TOKENS_117>|",
|
| 121 |
+
"|<EXTRA_TOKENS_118>|",
|
| 122 |
+
"|<EXTRA_TOKENS_119>|",
|
| 123 |
+
"|<EXTRA_TOKENS_120>|",
|
| 124 |
+
"|<EXTRA_TOKENS_121>|",
|
| 125 |
+
"|<EXTRA_TOKENS_122>|",
|
| 126 |
+
"|<EXTRA_TOKENS_123>|",
|
| 127 |
+
"|<EXTRA_TOKENS_124>|",
|
| 128 |
+
"|<EXTRA_TOKENS_125>|",
|
| 129 |
+
"|<EXTRA_TOKENS_126>|",
|
| 130 |
+
"|<EXTRA_TOKENS_127>|",
|
| 131 |
+
"|<EXTRA_TOKENS_128>|",
|
| 132 |
+
"|<EXTRA_TOKENS_129>|",
|
| 133 |
+
"|<EXTRA_TOKENS_130>|",
|
| 134 |
+
"|<EXTRA_TOKENS_131>|",
|
| 135 |
+
"|<EXTRA_TOKENS_132>|",
|
| 136 |
+
"|<EXTRA_TOKENS_133>|",
|
| 137 |
+
"|<EXTRA_TOKENS_134>|",
|
| 138 |
+
"|<EXTRA_TOKENS_135>|",
|
| 139 |
+
"|<EXTRA_TOKENS_136>|",
|
| 140 |
+
"|<EXTRA_TOKENS_137>|",
|
| 141 |
+
"|<EXTRA_TOKENS_138>|",
|
| 142 |
+
"|<EXTRA_TOKENS_139>|",
|
| 143 |
+
"|<EXTRA_TOKENS_140>|",
|
| 144 |
+
"|<EXTRA_TOKENS_141>|",
|
| 145 |
+
"|<EXTRA_TOKENS_142>|",
|
| 146 |
+
"|<EXTRA_TOKENS_143>|",
|
| 147 |
+
"|<EXTRA_TOKENS_144>|",
|
| 148 |
+
"|<EXTRA_TOKENS_145>|",
|
| 149 |
+
"|<EXTRA_TOKENS_146>|",
|
| 150 |
+
"|<EXTRA_TOKENS_147>|",
|
| 151 |
+
"|<EXTRA_TOKENS_148>|",
|
| 152 |
+
"|<EXTRA_TOKENS_149>|",
|
| 153 |
+
"|<EXTRA_TOKENS_150>|",
|
| 154 |
+
"|<EXTRA_TOKENS_151>|",
|
| 155 |
+
"|<EXTRA_TOKENS_152>|",
|
| 156 |
+
"|<EXTRA_TOKENS_153>|",
|
| 157 |
+
"|<EXTRA_TOKENS_154>|",
|
| 158 |
+
"|<EXTRA_TOKENS_155>|",
|
| 159 |
+
"|<EXTRA_TOKENS_156>|",
|
| 160 |
+
"|<EXTRA_TOKENS_157>|",
|
| 161 |
+
"|<EXTRA_TOKENS_158>|",
|
| 162 |
+
"|<EXTRA_TOKENS_159>|",
|
| 163 |
+
"|<EXTRA_TOKENS_160>|",
|
| 164 |
+
"|<EXTRA_TOKENS_161>|",
|
| 165 |
+
"|<EXTRA_TOKENS_162>|",
|
| 166 |
+
"|<EXTRA_TOKENS_163>|",
|
| 167 |
+
"|<EXTRA_TOKENS_164>|",
|
| 168 |
+
"|<EXTRA_TOKENS_165>|",
|
| 169 |
+
"|<EXTRA_TOKENS_166>|",
|
| 170 |
+
"|<EXTRA_TOKENS_167>|",
|
| 171 |
+
"|<EXTRA_TOKENS_168>|",
|
| 172 |
+
"|<EXTRA_TOKENS_169>|",
|
| 173 |
+
"|<EXTRA_TOKENS_170>|",
|
| 174 |
+
"|<EXTRA_TOKENS_171>|",
|
| 175 |
+
"|<EXTRA_TOKENS_172>|",
|
| 176 |
+
"|<EXTRA_TOKENS_173>|",
|
| 177 |
+
"|<EXTRA_TOKENS_174>|",
|
| 178 |
+
"|<EXTRA_TOKENS_175>|",
|
| 179 |
+
"|<EXTRA_TOKENS_176>|",
|
| 180 |
+
"|<EXTRA_TOKENS_177>|",
|
| 181 |
+
"|<EXTRA_TOKENS_178>|",
|
| 182 |
+
"|<EXTRA_TOKENS_179>|",
|
| 183 |
+
"|<EXTRA_TOKENS_180>|",
|
| 184 |
+
"|<EXTRA_TOKENS_181>|",
|
| 185 |
+
"|<EXTRA_TOKENS_182>|",
|
| 186 |
+
"|<EXTRA_TOKENS_183>|",
|
| 187 |
+
"|<EXTRA_TOKENS_184>|",
|
| 188 |
+
"|<EXTRA_TOKENS_185>|",
|
| 189 |
+
"|<EXTRA_TOKENS_186>|",
|
| 190 |
+
"|<EXTRA_TOKENS_187>|",
|
| 191 |
+
"|<EXTRA_TOKENS_188>|",
|
| 192 |
+
"|<EXTRA_TOKENS_189>|",
|
| 193 |
+
"|<EXTRA_TOKENS_190>|",
|
| 194 |
+
"|<EXTRA_TOKENS_191>|",
|
| 195 |
+
"|<EXTRA_TOKENS_192>|",
|
| 196 |
+
"|<EXTRA_TOKENS_193>|",
|
| 197 |
+
"|<EXTRA_TOKENS_194>|",
|
| 198 |
+
"|<EXTRA_TOKENS_195>|",
|
| 199 |
+
"|<EXTRA_TOKENS_196>|",
|
| 200 |
+
"|<EXTRA_TOKENS_197>|",
|
| 201 |
+
"|<EXTRA_TOKENS_198>|",
|
| 202 |
+
"|<EXTRA_TOKENS_199>|",
|
| 203 |
+
"|<EXTRA_TOKENS_200>|",
|
| 204 |
+
"|<EXTRA_TOKENS_201>|",
|
| 205 |
+
"|<EXTRA_TOKENS_202>|",
|
| 206 |
+
"|<EXTRA_TOKENS_203>|",
|
| 207 |
+
"|<EXTRA_TOKENS_204>|",
|
| 208 |
+
"|<EXTRA_TOKENS_205>|",
|
| 209 |
+
"|<EXTRA_TOKENS_206>|",
|
| 210 |
+
"|<EXTRA_TOKENS_207>|",
|
| 211 |
+
"|<EXTRA_TOKENS_208>|",
|
| 212 |
+
"|<EXTRA_TOKENS_209>|",
|
| 213 |
+
"|<EXTRA_TOKENS_210>|",
|
| 214 |
+
"|<EXTRA_TOKENS_211>|",
|
| 215 |
+
"|<EXTRA_TOKENS_212>|",
|
| 216 |
+
"|<EXTRA_TOKENS_213>|",
|
| 217 |
+
"|<EXTRA_TOKENS_214>|",
|
| 218 |
+
"|<EXTRA_TOKENS_215>|",
|
| 219 |
+
"|<EXTRA_TOKENS_216>|",
|
| 220 |
+
"|<EXTRA_TOKENS_217>|",
|
| 221 |
+
"|<EXTRA_TOKENS_218>|",
|
| 222 |
+
"|<EXTRA_TOKENS_219>|",
|
| 223 |
+
"|<EXTRA_TOKENS_220>|",
|
| 224 |
+
"|<EXTRA_TOKENS_221>|",
|
| 225 |
+
"|<EXTRA_TOKENS_222>|",
|
| 226 |
+
"|<EXTRA_TOKENS_223>|",
|
| 227 |
+
"|<EXTRA_TOKENS_224>|",
|
| 228 |
+
"|<EXTRA_TOKENS_225>|",
|
| 229 |
+
"|<EXTRA_TOKENS_226>|",
|
| 230 |
+
"|<EXTRA_TOKENS_227>|",
|
| 231 |
+
"|<EXTRA_TOKENS_228>|",
|
| 232 |
+
"|<EXTRA_TOKENS_229>|",
|
| 233 |
+
"|<EXTRA_TOKENS_230>|",
|
| 234 |
+
"|<EXTRA_TOKENS_231>|",
|
| 235 |
+
"|<EXTRA_TOKENS_232>|",
|
| 236 |
+
"|<EXTRA_TOKENS_233>|",
|
| 237 |
+
"|<EXTRA_TOKENS_234>|",
|
| 238 |
+
"|<EXTRA_TOKENS_235>|",
|
| 239 |
+
"|<EXTRA_TOKENS_236>|",
|
| 240 |
+
"|<EXTRA_TOKENS_237>|",
|
| 241 |
+
"|<EXTRA_TOKENS_238>|",
|
| 242 |
+
"|<EXTRA_TOKENS_239>|",
|
| 243 |
+
"|<EXTRA_TOKENS_240>|",
|
| 244 |
+
"|<EXTRA_TOKENS_241>|",
|
| 245 |
+
"|<EXTRA_TOKENS_242>|",
|
| 246 |
+
"|<EXTRA_TOKENS_243>|",
|
| 247 |
+
"|<EXTRA_TOKENS_244>|",
|
| 248 |
+
"|<EXTRA_TOKENS_245>|",
|
| 249 |
+
"|<EXTRA_TOKENS_246>|",
|
| 250 |
+
"|<EXTRA_TOKENS_247>|",
|
| 251 |
+
"|<EXTRA_TOKENS_248>|",
|
| 252 |
+
"|<EXTRA_TOKENS_249>|",
|
| 253 |
+
"|<EXTRA_TOKENS_250>|",
|
| 254 |
+
"|<EXTRA_TOKENS_251>|",
|
| 255 |
+
"|<EXTRA_TOKENS_252>|",
|
| 256 |
+
"|<EXTRA_TOKENS_253>|",
|
| 257 |
+
"|<EXTRA_TOKENS_254>|",
|
| 258 |
+
"|<EXTRA_TOKENS_255>|",
|
| 259 |
+
"|<EXTRA_TOKENS_256>|",
|
| 260 |
+
"|<EXTRA_TOKENS_257>|",
|
| 261 |
+
"|<EXTRA_TOKENS_258>|",
|
| 262 |
+
"|<EXTRA_TOKENS_259>|",
|
| 263 |
+
"|<EXTRA_TOKENS_260>|",
|
| 264 |
+
"|<EXTRA_TOKENS_261>|",
|
| 265 |
+
"|<EXTRA_TOKENS_262>|",
|
| 266 |
+
"|<EXTRA_TOKENS_263>|",
|
| 267 |
+
"|<EXTRA_TOKENS_264>|",
|
| 268 |
+
"|<EXTRA_TOKENS_265>|",
|
| 269 |
+
"|<EXTRA_TOKENS_266>|",
|
| 270 |
+
"<im_start>",
|
| 271 |
+
"<im_end>",
|
| 272 |
+
"<im_patch>",
|
| 273 |
+
"<im_col>",
|
| 274 |
+
"<low_res_im_start>",
|
| 275 |
+
"<|image|>",
|
| 276 |
+
"<im_low>",
|
| 277 |
+
"<frame_start>",
|
| 278 |
+
"<frame_end>",
|
| 279 |
+
"<|video|>"
|
| 280 |
+
],
|
| 281 |
+
"bos_token": "<|im_end|>",
|
| 282 |
+
"eos_token": {
|
| 283 |
+
"content": "<|im_end|>",
|
| 284 |
+
"lstrip": false,
|
| 285 |
+
"normalized": false,
|
| 286 |
+
"rstrip": false,
|
| 287 |
+
"single_word": false
|
| 288 |
+
},
|
| 289 |
+
"pad_token": {
|
| 290 |
+
"content": "<|endoftext|>",
|
| 291 |
+
"lstrip": false,
|
| 292 |
+
"normalized": false,
|
| 293 |
+
"rstrip": false,
|
| 294 |
+
"single_word": false
|
| 295 |
+
}
|
| 296 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,2726 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": false,
|
| 3 |
+
"add_prefix_space": false,
|
| 4 |
+
"added_tokens_decoder": {
|
| 5 |
+
"151643": {
|
| 6 |
+
"content": "<|endoftext|>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false,
|
| 11 |
+
"special": true
|
| 12 |
+
},
|
| 13 |
+
"151644": {
|
| 14 |
+
"content": "<|im_start|>",
|
| 15 |
+
"lstrip": false,
|
| 16 |
+
"normalized": false,
|
| 17 |
+
"rstrip": false,
|
| 18 |
+
"single_word": false,
|
| 19 |
+
"special": true
|
| 20 |
+
},
|
| 21 |
+
"151645": {
|
| 22 |
+
"content": "<|im_end|>",
|
| 23 |
+
"lstrip": false,
|
| 24 |
+
"normalized": false,
|
| 25 |
+
"rstrip": false,
|
| 26 |
+
"single_word": false,
|
| 27 |
+
"special": true
|
| 28 |
+
},
|
| 29 |
+
"151646": {
|
| 30 |
+
"content": "<|object_ref_start|>",
|
| 31 |
+
"lstrip": false,
|
| 32 |
+
"normalized": false,
|
| 33 |
+
"rstrip": false,
|
| 34 |
+
"single_word": false,
|
| 35 |
+
"special": true
|
| 36 |
+
},
|
| 37 |
+
"151647": {
|
| 38 |
+
"content": "<|object_ref_end|>",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": false,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false,
|
| 43 |
+
"special": true
|
| 44 |
+
},
|
| 45 |
+
"151648": {
|
| 46 |
+
"content": "<|box_start|>",
|
| 47 |
+
"lstrip": false,
|
| 48 |
+
"normalized": false,
|
| 49 |
+
"rstrip": false,
|
| 50 |
+
"single_word": false,
|
| 51 |
+
"special": true
|
| 52 |
+
},
|
| 53 |
+
"151649": {
|
| 54 |
+
"content": "<|box_end|>",
|
| 55 |
+
"lstrip": false,
|
| 56 |
+
"normalized": false,
|
| 57 |
+
"rstrip": false,
|
| 58 |
+
"single_word": false,
|
| 59 |
+
"special": true
|
| 60 |
+
},
|
| 61 |
+
"151650": {
|
| 62 |
+
"content": "<|quad_start|>",
|
| 63 |
+
"lstrip": false,
|
| 64 |
+
"normalized": false,
|
| 65 |
+
"rstrip": false,
|
| 66 |
+
"single_word": false,
|
| 67 |
+
"special": true
|
| 68 |
+
},
|
| 69 |
+
"151651": {
|
| 70 |
+
"content": "<|quad_end|>",
|
| 71 |
+
"lstrip": false,
|
| 72 |
+
"normalized": false,
|
| 73 |
+
"rstrip": false,
|
| 74 |
+
"single_word": false,
|
| 75 |
+
"special": true
|
| 76 |
+
},
|
| 77 |
+
"151652": {
|
| 78 |
+
"content": "<|vision_start|>",
|
| 79 |
+
"lstrip": false,
|
| 80 |
+
"normalized": false,
|
| 81 |
+
"rstrip": false,
|
| 82 |
+
"single_word": false,
|
| 83 |
+
"special": true
|
| 84 |
+
},
|
| 85 |
+
"151653": {
|
| 86 |
+
"content": "<|vision_end|>",
|
| 87 |
+
"lstrip": false,
|
| 88 |
+
"normalized": false,
|
| 89 |
+
"rstrip": false,
|
| 90 |
+
"single_word": false,
|
| 91 |
+
"special": true
|
| 92 |
+
},
|
| 93 |
+
"151654": {
|
| 94 |
+
"content": "<|vision_pad|>",
|
| 95 |
+
"lstrip": false,
|
| 96 |
+
"normalized": false,
|
| 97 |
+
"rstrip": false,
|
| 98 |
+
"single_word": false,
|
| 99 |
+
"special": true
|
| 100 |
+
},
|
| 101 |
+
"151655": {
|
| 102 |
+
"content": "<|image_pad|>",
|
| 103 |
+
"lstrip": false,
|
| 104 |
+
"normalized": false,
|
| 105 |
+
"rstrip": false,
|
| 106 |
+
"single_word": false,
|
| 107 |
+
"special": true
|
| 108 |
+
},
|
| 109 |
+
"151656": {
|
| 110 |
+
"content": "<|video_pad|>",
|
| 111 |
+
"lstrip": false,
|
| 112 |
+
"normalized": false,
|
| 113 |
+
"rstrip": false,
|
| 114 |
+
"single_word": false,
|
| 115 |
+
"special": true
|
| 116 |
+
},
|
| 117 |
+
"151657": {
|
| 118 |
+
"content": "<tool_call>",
|
| 119 |
+
"lstrip": false,
|
| 120 |
+
"normalized": false,
|
| 121 |
+
"rstrip": false,
|
| 122 |
+
"single_word": false,
|
| 123 |
+
"special": false
|
| 124 |
+
},
|
| 125 |
+
"151658": {
|
| 126 |
+
"content": "</tool_call>",
|
| 127 |
+
"lstrip": false,
|
| 128 |
+
"normalized": false,
|
| 129 |
+
"rstrip": false,
|
| 130 |
+
"single_word": false,
|
| 131 |
+
"special": false
|
| 132 |
+
},
|
| 133 |
+
"151659": {
|
| 134 |
+
"content": "<|fim_prefix|>",
|
| 135 |
+
"lstrip": false,
|
| 136 |
+
"normalized": false,
|
| 137 |
+
"rstrip": false,
|
| 138 |
+
"single_word": false,
|
| 139 |
+
"special": false
|
| 140 |
+
},
|
| 141 |
+
"151660": {
|
| 142 |
+
"content": "<|fim_middle|>",
|
| 143 |
+
"lstrip": false,
|
| 144 |
+
"normalized": false,
|
| 145 |
+
"rstrip": false,
|
| 146 |
+
"single_word": false,
|
| 147 |
+
"special": false
|
| 148 |
+
},
|
| 149 |
+
"151661": {
|
| 150 |
+
"content": "<|fim_suffix|>",
|
| 151 |
+
"lstrip": false,
|
| 152 |
+
"normalized": false,
|
| 153 |
+
"rstrip": false,
|
| 154 |
+
"single_word": false,
|
| 155 |
+
"special": false
|
| 156 |
+
},
|
| 157 |
+
"151662": {
|
| 158 |
+
"content": "<|fim_pad|>",
|
| 159 |
+
"lstrip": false,
|
| 160 |
+
"normalized": false,
|
| 161 |
+
"rstrip": false,
|
| 162 |
+
"single_word": false,
|
| 163 |
+
"special": false
|
| 164 |
+
},
|
| 165 |
+
"151663": {
|
| 166 |
+
"content": "<|repo_name|>",
|
| 167 |
+
"lstrip": false,
|
| 168 |
+
"normalized": false,
|
| 169 |
+
"rstrip": false,
|
| 170 |
+
"single_word": false,
|
| 171 |
+
"special": false
|
| 172 |
+
},
|
| 173 |
+
"151664": {
|
| 174 |
+
"content": "<|file_sep|>",
|
| 175 |
+
"lstrip": false,
|
| 176 |
+
"normalized": false,
|
| 177 |
+
"rstrip": false,
|
| 178 |
+
"single_word": false,
|
| 179 |
+
"special": false
|
| 180 |
+
},
|
| 181 |
+
"151665": {
|
| 182 |
+
"content": "<tool_response>",
|
| 183 |
+
"lstrip": false,
|
| 184 |
+
"normalized": false,
|
| 185 |
+
"rstrip": false,
|
| 186 |
+
"single_word": false,
|
| 187 |
+
"special": false
|
| 188 |
+
},
|
| 189 |
+
"151666": {
|
| 190 |
+
"content": "</tool_response>",
|
| 191 |
+
"lstrip": false,
|
| 192 |
+
"normalized": false,
|
| 193 |
+
"rstrip": false,
|
| 194 |
+
"single_word": false,
|
| 195 |
+
"special": false
|
| 196 |
+
},
|
| 197 |
+
"151667": {
|
| 198 |
+
"content": "<think>",
|
| 199 |
+
"lstrip": false,
|
| 200 |
+
"normalized": false,
|
| 201 |
+
"rstrip": false,
|
| 202 |
+
"single_word": false,
|
| 203 |
+
"special": false
|
| 204 |
+
},
|
| 205 |
+
"151668": {
|
| 206 |
+
"content": "</think>",
|
| 207 |
+
"lstrip": false,
|
| 208 |
+
"normalized": false,
|
| 209 |
+
"rstrip": false,
|
| 210 |
+
"single_word": false,
|
| 211 |
+
"special": false
|
| 212 |
+
},
|
| 213 |
+
"151669": {
|
| 214 |
+
"content": "|<EXTRA_TOKENS_0>|",
|
| 215 |
+
"lstrip": false,
|
| 216 |
+
"normalized": false,
|
| 217 |
+
"rstrip": false,
|
| 218 |
+
"single_word": false,
|
| 219 |
+
"special": true
|
| 220 |
+
},
|
| 221 |
+
"151670": {
|
| 222 |
+
"content": "|<EXTRA_TOKENS_1>|",
|
| 223 |
+
"lstrip": false,
|
| 224 |
+
"normalized": false,
|
| 225 |
+
"rstrip": false,
|
| 226 |
+
"single_word": false,
|
| 227 |
+
"special": true
|
| 228 |
+
},
|
| 229 |
+
"151671": {
|
| 230 |
+
"content": "|<EXTRA_TOKENS_2>|",
|
| 231 |
+
"lstrip": false,
|
| 232 |
+
"normalized": false,
|
| 233 |
+
"rstrip": false,
|
| 234 |
+
"single_word": false,
|
| 235 |
+
"special": true
|
| 236 |
+
},
|
| 237 |
+
"151672": {
|
| 238 |
+
"content": "|<EXTRA_TOKENS_3>|",
|
| 239 |
+
"lstrip": false,
|
| 240 |
+
"normalized": false,
|
| 241 |
+
"rstrip": false,
|
| 242 |
+
"single_word": false,
|
| 243 |
+
"special": true
|
| 244 |
+
},
|
| 245 |
+
"151673": {
|
| 246 |
+
"content": "|<EXTRA_TOKENS_4>|",
|
| 247 |
+
"lstrip": false,
|
| 248 |
+
"normalized": false,
|
| 249 |
+
"rstrip": false,
|
| 250 |
+
"single_word": false,
|
| 251 |
+
"special": true
|
| 252 |
+
},
|
| 253 |
+
"151674": {
|
| 254 |
+
"content": "|<EXTRA_TOKENS_5>|",
|
| 255 |
+
"lstrip": false,
|
| 256 |
+
"normalized": false,
|
| 257 |
+
"rstrip": false,
|
| 258 |
+
"single_word": false,
|
| 259 |
+
"special": true
|
| 260 |
+
},
|
| 261 |
+
"151675": {
|
| 262 |
+
"content": "|<EXTRA_TOKENS_6>|",
|
| 263 |
+
"lstrip": false,
|
| 264 |
+
"normalized": false,
|
| 265 |
+
"rstrip": false,
|
| 266 |
+
"single_word": false,
|
| 267 |
+
"special": true
|
| 268 |
+
},
|
| 269 |
+
"151676": {
|
| 270 |
+
"content": "|<EXTRA_TOKENS_7>|",
|
| 271 |
+
"lstrip": false,
|
| 272 |
+
"normalized": false,
|
| 273 |
+
"rstrip": false,
|
| 274 |
+
"single_word": false,
|
| 275 |
+
"special": true
|
| 276 |
+
},
|
| 277 |
+
"151677": {
|
| 278 |
+
"content": "|<EXTRA_TOKENS_8>|",
|
| 279 |
+
"lstrip": false,
|
| 280 |
+
"normalized": false,
|
| 281 |
+
"rstrip": false,
|
| 282 |
+
"single_word": false,
|
| 283 |
+
"special": true
|
| 284 |
+
},
|
| 285 |
+
"151678": {
|
| 286 |
+
"content": "|<EXTRA_TOKENS_9>|",
|
| 287 |
+
"lstrip": false,
|
| 288 |
+
"normalized": false,
|
| 289 |
+
"rstrip": false,
|
| 290 |
+
"single_word": false,
|
| 291 |
+
"special": true
|
| 292 |
+
},
|
| 293 |
+
"151679": {
|
| 294 |
+
"content": "|<EXTRA_TOKENS_10>|",
|
| 295 |
+
"lstrip": false,
|
| 296 |
+
"normalized": false,
|
| 297 |
+
"rstrip": false,
|
| 298 |
+
"single_word": false,
|
| 299 |
+
"special": true
|
| 300 |
+
},
|
| 301 |
+
"151680": {
|
| 302 |
+
"content": "|<EXTRA_TOKENS_11>|",
|
| 303 |
+
"lstrip": false,
|
| 304 |
+
"normalized": false,
|
| 305 |
+
"rstrip": false,
|
| 306 |
+
"single_word": false,
|
| 307 |
+
"special": true
|
| 308 |
+
},
|
| 309 |
+
"151681": {
|
| 310 |
+
"content": "|<EXTRA_TOKENS_12>|",
|
| 311 |
+
"lstrip": false,
|
| 312 |
+
"normalized": false,
|
| 313 |
+
"rstrip": false,
|
| 314 |
+
"single_word": false,
|
| 315 |
+
"special": true
|
| 316 |
+
},
|
| 317 |
+
"151682": {
|
| 318 |
+
"content": "|<EXTRA_TOKENS_13>|",
|
| 319 |
+
"lstrip": false,
|
| 320 |
+
"normalized": false,
|
| 321 |
+
"rstrip": false,
|
| 322 |
+
"single_word": false,
|
| 323 |
+
"special": true
|
| 324 |
+
},
|
| 325 |
+
"151683": {
|
| 326 |
+
"content": "|<EXTRA_TOKENS_14>|",
|
| 327 |
+
"lstrip": false,
|
| 328 |
+
"normalized": false,
|
| 329 |
+
"rstrip": false,
|
| 330 |
+
"single_word": false,
|
| 331 |
+
"special": true
|
| 332 |
+
},
|
| 333 |
+
"151684": {
|
| 334 |
+
"content": "|<EXTRA_TOKENS_15>|",
|
| 335 |
+
"lstrip": false,
|
| 336 |
+
"normalized": false,
|
| 337 |
+
"rstrip": false,
|
| 338 |
+
"single_word": false,
|
| 339 |
+
"special": true
|
| 340 |
+
},
|
| 341 |
+
"151685": {
|
| 342 |
+
"content": "|<EXTRA_TOKENS_16>|",
|
| 343 |
+
"lstrip": false,
|
| 344 |
+
"normalized": false,
|
| 345 |
+
"rstrip": false,
|
| 346 |
+
"single_word": false,
|
| 347 |
+
"special": true
|
| 348 |
+
},
|
| 349 |
+
"151686": {
|
| 350 |
+
"content": "|<EXTRA_TOKENS_17>|",
|
| 351 |
+
"lstrip": false,
|
| 352 |
+
"normalized": false,
|
| 353 |
+
"rstrip": false,
|
| 354 |
+
"single_word": false,
|
| 355 |
+
"special": true
|
| 356 |
+
},
|
| 357 |
+
"151687": {
|
| 358 |
+
"content": "|<EXTRA_TOKENS_18>|",
|
| 359 |
+
"lstrip": false,
|
| 360 |
+
"normalized": false,
|
| 361 |
+
"rstrip": false,
|
| 362 |
+
"single_word": false,
|
| 363 |
+
"special": true
|
| 364 |
+
},
|
| 365 |
+
"151688": {
|
| 366 |
+
"content": "|<EXTRA_TOKENS_19>|",
|
| 367 |
+
"lstrip": false,
|
| 368 |
+
"normalized": false,
|
| 369 |
+
"rstrip": false,
|
| 370 |
+
"single_word": false,
|
| 371 |
+
"special": true
|
| 372 |
+
},
|
| 373 |
+
"151689": {
|
| 374 |
+
"content": "|<EXTRA_TOKENS_20>|",
|
| 375 |
+
"lstrip": false,
|
| 376 |
+
"normalized": false,
|
| 377 |
+
"rstrip": false,
|
| 378 |
+
"single_word": false,
|
| 379 |
+
"special": true
|
| 380 |
+
},
|
| 381 |
+
"151690": {
|
| 382 |
+
"content": "|<EXTRA_TOKENS_21>|",
|
| 383 |
+
"lstrip": false,
|
| 384 |
+
"normalized": false,
|
| 385 |
+
"rstrip": false,
|
| 386 |
+
"single_word": false,
|
| 387 |
+
"special": true
|
| 388 |
+
},
|
| 389 |
+
"151691": {
|
| 390 |
+
"content": "|<EXTRA_TOKENS_22>|",
|
| 391 |
+
"lstrip": false,
|
| 392 |
+
"normalized": false,
|
| 393 |
+
"rstrip": false,
|
| 394 |
+
"single_word": false,
|
| 395 |
+
"special": true
|
| 396 |
+
},
|
| 397 |
+
"151692": {
|
| 398 |
+
"content": "|<EXTRA_TOKENS_23>|",
|
| 399 |
+
"lstrip": false,
|
| 400 |
+
"normalized": false,
|
| 401 |
+
"rstrip": false,
|
| 402 |
+
"single_word": false,
|
| 403 |
+
"special": true
|
| 404 |
+
},
|
| 405 |
+
"151693": {
|
| 406 |
+
"content": "|<EXTRA_TOKENS_24>|",
|
| 407 |
+
"lstrip": false,
|
| 408 |
+
"normalized": false,
|
| 409 |
+
"rstrip": false,
|
| 410 |
+
"single_word": false,
|
| 411 |
+
"special": true
|
| 412 |
+
},
|
| 413 |
+
"151694": {
|
| 414 |
+
"content": "|<EXTRA_TOKENS_25>|",
|
| 415 |
+
"lstrip": false,
|
| 416 |
+
"normalized": false,
|
| 417 |
+
"rstrip": false,
|
| 418 |
+
"single_word": false,
|
| 419 |
+
"special": true
|
| 420 |
+
},
|
| 421 |
+
"151695": {
|
| 422 |
+
"content": "|<EXTRA_TOKENS_26>|",
|
| 423 |
+
"lstrip": false,
|
| 424 |
+
"normalized": false,
|
| 425 |
+
"rstrip": false,
|
| 426 |
+
"single_word": false,
|
| 427 |
+
"special": true
|
| 428 |
+
},
|
| 429 |
+
"151696": {
|
| 430 |
+
"content": "|<EXTRA_TOKENS_27>|",
|
| 431 |
+
"lstrip": false,
|
| 432 |
+
"normalized": false,
|
| 433 |
+
"rstrip": false,
|
| 434 |
+
"single_word": false,
|
| 435 |
+
"special": true
|
| 436 |
+
},
|
| 437 |
+
"151697": {
|
| 438 |
+
"content": "|<EXTRA_TOKENS_28>|",
|
| 439 |
+
"lstrip": false,
|
| 440 |
+
"normalized": false,
|
| 441 |
+
"rstrip": false,
|
| 442 |
+
"single_word": false,
|
| 443 |
+
"special": true
|
| 444 |
+
},
|
| 445 |
+
"151698": {
|
| 446 |
+
"content": "|<EXTRA_TOKENS_29>|",
|
| 447 |
+
"lstrip": false,
|
| 448 |
+
"normalized": false,
|
| 449 |
+
"rstrip": false,
|
| 450 |
+
"single_word": false,
|
| 451 |
+
"special": true
|
| 452 |
+
},
|
| 453 |
+
"151699": {
|
| 454 |
+
"content": "|<EXTRA_TOKENS_30>|",
|
| 455 |
+
"lstrip": false,
|
| 456 |
+
"normalized": false,
|
| 457 |
+
"rstrip": false,
|
| 458 |
+
"single_word": false,
|
| 459 |
+
"special": true
|
| 460 |
+
},
|
| 461 |
+
"151700": {
|
| 462 |
+
"content": "|<EXTRA_TOKENS_31>|",
|
| 463 |
+
"lstrip": false,
|
| 464 |
+
"normalized": false,
|
| 465 |
+
"rstrip": false,
|
| 466 |
+
"single_word": false,
|
| 467 |
+
"special": true
|
| 468 |
+
},
|
| 469 |
+
"151701": {
|
| 470 |
+
"content": "|<EXTRA_TOKENS_32>|",
|
| 471 |
+
"lstrip": false,
|
| 472 |
+
"normalized": false,
|
| 473 |
+
"rstrip": false,
|
| 474 |
+
"single_word": false,
|
| 475 |
+
"special": true
|
| 476 |
+
},
|
| 477 |
+
"151702": {
|
| 478 |
+
"content": "|<EXTRA_TOKENS_33>|",
|
| 479 |
+
"lstrip": false,
|
| 480 |
+
"normalized": false,
|
| 481 |
+
"rstrip": false,
|
| 482 |
+
"single_word": false,
|
| 483 |
+
"special": true
|
| 484 |
+
},
|
| 485 |
+
"151703": {
|
| 486 |
+
"content": "|<EXTRA_TOKENS_34>|",
|
| 487 |
+
"lstrip": false,
|
| 488 |
+
"normalized": false,
|
| 489 |
+
"rstrip": false,
|
| 490 |
+
"single_word": false,
|
| 491 |
+
"special": true
|
| 492 |
+
},
|
| 493 |
+
"151704": {
|
| 494 |
+
"content": "|<EXTRA_TOKENS_35>|",
|
| 495 |
+
"lstrip": false,
|
| 496 |
+
"normalized": false,
|
| 497 |
+
"rstrip": false,
|
| 498 |
+
"single_word": false,
|
| 499 |
+
"special": true
|
| 500 |
+
},
|
| 501 |
+
"151705": {
|
| 502 |
+
"content": "|<EXTRA_TOKENS_36>|",
|
| 503 |
+
"lstrip": false,
|
| 504 |
+
"normalized": false,
|
| 505 |
+
"rstrip": false,
|
| 506 |
+
"single_word": false,
|
| 507 |
+
"special": true
|
| 508 |
+
},
|
| 509 |
+
"151706": {
|
| 510 |
+
"content": "|<EXTRA_TOKENS_37>|",
|
| 511 |
+
"lstrip": false,
|
| 512 |
+
"normalized": false,
|
| 513 |
+
"rstrip": false,
|
| 514 |
+
"single_word": false,
|
| 515 |
+
"special": true
|
| 516 |
+
},
|
| 517 |
+
"151707": {
|
| 518 |
+
"content": "|<EXTRA_TOKENS_38>|",
|
| 519 |
+
"lstrip": false,
|
| 520 |
+
"normalized": false,
|
| 521 |
+
"rstrip": false,
|
| 522 |
+
"single_word": false,
|
| 523 |
+
"special": true
|
| 524 |
+
},
|
| 525 |
+
"151708": {
|
| 526 |
+
"content": "|<EXTRA_TOKENS_39>|",
|
| 527 |
+
"lstrip": false,
|
| 528 |
+
"normalized": false,
|
| 529 |
+
"rstrip": false,
|
| 530 |
+
"single_word": false,
|
| 531 |
+
"special": true
|
| 532 |
+
},
|
| 533 |
+
"151709": {
|
| 534 |
+
"content": "|<EXTRA_TOKENS_40>|",
|
| 535 |
+
"lstrip": false,
|
| 536 |
+
"normalized": false,
|
| 537 |
+
"rstrip": false,
|
| 538 |
+
"single_word": false,
|
| 539 |
+
"special": true
|
| 540 |
+
},
|
| 541 |
+
"151710": {
|
| 542 |
+
"content": "|<EXTRA_TOKENS_41>|",
|
| 543 |
+
"lstrip": false,
|
| 544 |
+
"normalized": false,
|
| 545 |
+
"rstrip": false,
|
| 546 |
+
"single_word": false,
|
| 547 |
+
"special": true
|
| 548 |
+
},
|
| 549 |
+
"151711": {
|
| 550 |
+
"content": "|<EXTRA_TOKENS_42>|",
|
| 551 |
+
"lstrip": false,
|
| 552 |
+
"normalized": false,
|
| 553 |
+
"rstrip": false,
|
| 554 |
+
"single_word": false,
|
| 555 |
+
"special": true
|
| 556 |
+
},
|
| 557 |
+
"151712": {
|
| 558 |
+
"content": "|<EXTRA_TOKENS_43>|",
|
| 559 |
+
"lstrip": false,
|
| 560 |
+
"normalized": false,
|
| 561 |
+
"rstrip": false,
|
| 562 |
+
"single_word": false,
|
| 563 |
+
"special": true
|
| 564 |
+
},
|
| 565 |
+
"151713": {
|
| 566 |
+
"content": "|<EXTRA_TOKENS_44>|",
|
| 567 |
+
"lstrip": false,
|
| 568 |
+
"normalized": false,
|
| 569 |
+
"rstrip": false,
|
| 570 |
+
"single_word": false,
|
| 571 |
+
"special": true
|
| 572 |
+
},
|
| 573 |
+
"151714": {
|
| 574 |
+
"content": "|<EXTRA_TOKENS_45>|",
|
| 575 |
+
"lstrip": false,
|
| 576 |
+
"normalized": false,
|
| 577 |
+
"rstrip": false,
|
| 578 |
+
"single_word": false,
|
| 579 |
+
"special": true
|
| 580 |
+
},
|
| 581 |
+
"151715": {
|
| 582 |
+
"content": "|<EXTRA_TOKENS_46>|",
|
| 583 |
+
"lstrip": false,
|
| 584 |
+
"normalized": false,
|
| 585 |
+
"rstrip": false,
|
| 586 |
+
"single_word": false,
|
| 587 |
+
"special": true
|
| 588 |
+
},
|
| 589 |
+
"151716": {
|
| 590 |
+
"content": "|<EXTRA_TOKENS_47>|",
|
| 591 |
+
"lstrip": false,
|
| 592 |
+
"normalized": false,
|
| 593 |
+
"rstrip": false,
|
| 594 |
+
"single_word": false,
|
| 595 |
+
"special": true
|
| 596 |
+
},
|
| 597 |
+
"151717": {
|
| 598 |
+
"content": "|<EXTRA_TOKENS_48>|",
|
| 599 |
+
"lstrip": false,
|
| 600 |
+
"normalized": false,
|
| 601 |
+
"rstrip": false,
|
| 602 |
+
"single_word": false,
|
| 603 |
+
"special": true
|
| 604 |
+
},
|
| 605 |
+
"151718": {
|
| 606 |
+
"content": "|<EXTRA_TOKENS_49>|",
|
| 607 |
+
"lstrip": false,
|
| 608 |
+
"normalized": false,
|
| 609 |
+
"rstrip": false,
|
| 610 |
+
"single_word": false,
|
| 611 |
+
"special": true
|
| 612 |
+
},
|
| 613 |
+
"151719": {
|
| 614 |
+
"content": "|<EXTRA_TOKENS_50>|",
|
| 615 |
+
"lstrip": false,
|
| 616 |
+
"normalized": false,
|
| 617 |
+
"rstrip": false,
|
| 618 |
+
"single_word": false,
|
| 619 |
+
"special": true
|
| 620 |
+
},
|
| 621 |
+
"151720": {
|
| 622 |
+
"content": "|<EXTRA_TOKENS_51>|",
|
| 623 |
+
"lstrip": false,
|
| 624 |
+
"normalized": false,
|
| 625 |
+
"rstrip": false,
|
| 626 |
+
"single_word": false,
|
| 627 |
+
"special": true
|
| 628 |
+
},
|
| 629 |
+
"151721": {
|
| 630 |
+
"content": "|<EXTRA_TOKENS_52>|",
|
| 631 |
+
"lstrip": false,
|
| 632 |
+
"normalized": false,
|
| 633 |
+
"rstrip": false,
|
| 634 |
+
"single_word": false,
|
| 635 |
+
"special": true
|
| 636 |
+
},
|
| 637 |
+
"151722": {
|
| 638 |
+
"content": "|<EXTRA_TOKENS_53>|",
|
| 639 |
+
"lstrip": false,
|
| 640 |
+
"normalized": false,
|
| 641 |
+
"rstrip": false,
|
| 642 |
+
"single_word": false,
|
| 643 |
+
"special": true
|
| 644 |
+
},
|
| 645 |
+
"151723": {
|
| 646 |
+
"content": "|<EXTRA_TOKENS_54>|",
|
| 647 |
+
"lstrip": false,
|
| 648 |
+
"normalized": false,
|
| 649 |
+
"rstrip": false,
|
| 650 |
+
"single_word": false,
|
| 651 |
+
"special": true
|
| 652 |
+
},
|
| 653 |
+
"151724": {
|
| 654 |
+
"content": "|<EXTRA_TOKENS_55>|",
|
| 655 |
+
"lstrip": false,
|
| 656 |
+
"normalized": false,
|
| 657 |
+
"rstrip": false,
|
| 658 |
+
"single_word": false,
|
| 659 |
+
"special": true
|
| 660 |
+
},
|
| 661 |
+
"151725": {
|
| 662 |
+
"content": "|<EXTRA_TOKENS_56>|",
|
| 663 |
+
"lstrip": false,
|
| 664 |
+
"normalized": false,
|
| 665 |
+
"rstrip": false,
|
| 666 |
+
"single_word": false,
|
| 667 |
+
"special": true
|
| 668 |
+
},
|
| 669 |
+
"151726": {
|
| 670 |
+
"content": "|<EXTRA_TOKENS_57>|",
|
| 671 |
+
"lstrip": false,
|
| 672 |
+
"normalized": false,
|
| 673 |
+
"rstrip": false,
|
| 674 |
+
"single_word": false,
|
| 675 |
+
"special": true
|
| 676 |
+
},
|
| 677 |
+
"151727": {
|
| 678 |
+
"content": "|<EXTRA_TOKENS_58>|",
|
| 679 |
+
"lstrip": false,
|
| 680 |
+
"normalized": false,
|
| 681 |
+
"rstrip": false,
|
| 682 |
+
"single_word": false,
|
| 683 |
+
"special": true
|
| 684 |
+
},
|
| 685 |
+
"151728": {
|
| 686 |
+
"content": "|<EXTRA_TOKENS_59>|",
|
| 687 |
+
"lstrip": false,
|
| 688 |
+
"normalized": false,
|
| 689 |
+
"rstrip": false,
|
| 690 |
+
"single_word": false,
|
| 691 |
+
"special": true
|
| 692 |
+
},
|
| 693 |
+
"151729": {
|
| 694 |
+
"content": "|<EXTRA_TOKENS_60>|",
|
| 695 |
+
"lstrip": false,
|
| 696 |
+
"normalized": false,
|
| 697 |
+
"rstrip": false,
|
| 698 |
+
"single_word": false,
|
| 699 |
+
"special": true
|
| 700 |
+
},
|
| 701 |
+
"151730": {
|
| 702 |
+
"content": "|<EXTRA_TOKENS_61>|",
|
| 703 |
+
"lstrip": false,
|
| 704 |
+
"normalized": false,
|
| 705 |
+
"rstrip": false,
|
| 706 |
+
"single_word": false,
|
| 707 |
+
"special": true
|
| 708 |
+
},
|
| 709 |
+
"151731": {
|
| 710 |
+
"content": "|<EXTRA_TOKENS_62>|",
|
| 711 |
+
"lstrip": false,
|
| 712 |
+
"normalized": false,
|
| 713 |
+
"rstrip": false,
|
| 714 |
+
"single_word": false,
|
| 715 |
+
"special": true
|
| 716 |
+
},
|
| 717 |
+
"151732": {
|
| 718 |
+
"content": "|<EXTRA_TOKENS_63>|",
|
| 719 |
+
"lstrip": false,
|
| 720 |
+
"normalized": false,
|
| 721 |
+
"rstrip": false,
|
| 722 |
+
"single_word": false,
|
| 723 |
+
"special": true
|
| 724 |
+
},
|
| 725 |
+
"151733": {
|
| 726 |
+
"content": "|<EXTRA_TOKENS_64>|",
|
| 727 |
+
"lstrip": false,
|
| 728 |
+
"normalized": false,
|
| 729 |
+
"rstrip": false,
|
| 730 |
+
"single_word": false,
|
| 731 |
+
"special": true
|
| 732 |
+
},
|
| 733 |
+
"151734": {
|
| 734 |
+
"content": "|<EXTRA_TOKENS_65>|",
|
| 735 |
+
"lstrip": false,
|
| 736 |
+
"normalized": false,
|
| 737 |
+
"rstrip": false,
|
| 738 |
+
"single_word": false,
|
| 739 |
+
"special": true
|
| 740 |
+
},
|
| 741 |
+
"151735": {
|
| 742 |
+
"content": "|<EXTRA_TOKENS_66>|",
|
| 743 |
+
"lstrip": false,
|
| 744 |
+
"normalized": false,
|
| 745 |
+
"rstrip": false,
|
| 746 |
+
"single_word": false,
|
| 747 |
+
"special": true
|
| 748 |
+
},
|
| 749 |
+
"151736": {
|
| 750 |
+
"content": "|<EXTRA_TOKENS_67>|",
|
| 751 |
+
"lstrip": false,
|
| 752 |
+
"normalized": false,
|
| 753 |
+
"rstrip": false,
|
| 754 |
+
"single_word": false,
|
| 755 |
+
"special": true
|
| 756 |
+
},
|
| 757 |
+
"151737": {
|
| 758 |
+
"content": "|<EXTRA_TOKENS_68>|",
|
| 759 |
+
"lstrip": false,
|
| 760 |
+
"normalized": false,
|
| 761 |
+
"rstrip": false,
|
| 762 |
+
"single_word": false,
|
| 763 |
+
"special": true
|
| 764 |
+
},
|
| 765 |
+
"151738": {
|
| 766 |
+
"content": "|<EXTRA_TOKENS_69>|",
|
| 767 |
+
"lstrip": false,
|
| 768 |
+
"normalized": false,
|
| 769 |
+
"rstrip": false,
|
| 770 |
+
"single_word": false,
|
| 771 |
+
"special": true
|
| 772 |
+
},
|
| 773 |
+
"151739": {
|
| 774 |
+
"content": "|<EXTRA_TOKENS_70>|",
|
| 775 |
+
"lstrip": false,
|
| 776 |
+
"normalized": false,
|
| 777 |
+
"rstrip": false,
|
| 778 |
+
"single_word": false,
|
| 779 |
+
"special": true
|
| 780 |
+
},
|
| 781 |
+
"151740": {
|
| 782 |
+
"content": "|<EXTRA_TOKENS_71>|",
|
| 783 |
+
"lstrip": false,
|
| 784 |
+
"normalized": false,
|
| 785 |
+
"rstrip": false,
|
| 786 |
+
"single_word": false,
|
| 787 |
+
"special": true
|
| 788 |
+
},
|
| 789 |
+
"151741": {
|
| 790 |
+
"content": "|<EXTRA_TOKENS_72>|",
|
| 791 |
+
"lstrip": false,
|
| 792 |
+
"normalized": false,
|
| 793 |
+
"rstrip": false,
|
| 794 |
+
"single_word": false,
|
| 795 |
+
"special": true
|
| 796 |
+
},
|
| 797 |
+
"151742": {
|
| 798 |
+
"content": "|<EXTRA_TOKENS_73>|",
|
| 799 |
+
"lstrip": false,
|
| 800 |
+
"normalized": false,
|
| 801 |
+
"rstrip": false,
|
| 802 |
+
"single_word": false,
|
| 803 |
+
"special": true
|
| 804 |
+
},
|
| 805 |
+
"151743": {
|
| 806 |
+
"content": "|<EXTRA_TOKENS_74>|",
|
| 807 |
+
"lstrip": false,
|
| 808 |
+
"normalized": false,
|
| 809 |
+
"rstrip": false,
|
| 810 |
+
"single_word": false,
|
| 811 |
+
"special": true
|
| 812 |
+
},
|
| 813 |
+
"151744": {
|
| 814 |
+
"content": "|<EXTRA_TOKENS_75>|",
|
| 815 |
+
"lstrip": false,
|
| 816 |
+
"normalized": false,
|
| 817 |
+
"rstrip": false,
|
| 818 |
+
"single_word": false,
|
| 819 |
+
"special": true
|
| 820 |
+
},
|
| 821 |
+
"151745": {
|
| 822 |
+
"content": "|<EXTRA_TOKENS_76>|",
|
| 823 |
+
"lstrip": false,
|
| 824 |
+
"normalized": false,
|
| 825 |
+
"rstrip": false,
|
| 826 |
+
"single_word": false,
|
| 827 |
+
"special": true
|
| 828 |
+
},
|
| 829 |
+
"151746": {
|
| 830 |
+
"content": "|<EXTRA_TOKENS_77>|",
|
| 831 |
+
"lstrip": false,
|
| 832 |
+
"normalized": false,
|
| 833 |
+
"rstrip": false,
|
| 834 |
+
"single_word": false,
|
| 835 |
+
"special": true
|
| 836 |
+
},
|
| 837 |
+
"151747": {
|
| 838 |
+
"content": "|<EXTRA_TOKENS_78>|",
|
| 839 |
+
"lstrip": false,
|
| 840 |
+
"normalized": false,
|
| 841 |
+
"rstrip": false,
|
| 842 |
+
"single_word": false,
|
| 843 |
+
"special": true
|
| 844 |
+
},
|
| 845 |
+
"151748": {
|
| 846 |
+
"content": "|<EXTRA_TOKENS_79>|",
|
| 847 |
+
"lstrip": false,
|
| 848 |
+
"normalized": false,
|
| 849 |
+
"rstrip": false,
|
| 850 |
+
"single_word": false,
|
| 851 |
+
"special": true
|
| 852 |
+
},
|
| 853 |
+
"151749": {
|
| 854 |
+
"content": "|<EXTRA_TOKENS_80>|",
|
| 855 |
+
"lstrip": false,
|
| 856 |
+
"normalized": false,
|
| 857 |
+
"rstrip": false,
|
| 858 |
+
"single_word": false,
|
| 859 |
+
"special": true
|
| 860 |
+
},
|
| 861 |
+
"151750": {
|
| 862 |
+
"content": "|<EXTRA_TOKENS_81>|",
|
| 863 |
+
"lstrip": false,
|
| 864 |
+
"normalized": false,
|
| 865 |
+
"rstrip": false,
|
| 866 |
+
"single_word": false,
|
| 867 |
+
"special": true
|
| 868 |
+
},
|
| 869 |
+
"151751": {
|
| 870 |
+
"content": "|<EXTRA_TOKENS_82>|",
|
| 871 |
+
"lstrip": false,
|
| 872 |
+
"normalized": false,
|
| 873 |
+
"rstrip": false,
|
| 874 |
+
"single_word": false,
|
| 875 |
+
"special": true
|
| 876 |
+
},
|
| 877 |
+
"151752": {
|
| 878 |
+
"content": "|<EXTRA_TOKENS_83>|",
|
| 879 |
+
"lstrip": false,
|
| 880 |
+
"normalized": false,
|
| 881 |
+
"rstrip": false,
|
| 882 |
+
"single_word": false,
|
| 883 |
+
"special": true
|
| 884 |
+
},
|
| 885 |
+
"151753": {
|
| 886 |
+
"content": "|<EXTRA_TOKENS_84>|",
|
| 887 |
+
"lstrip": false,
|
| 888 |
+
"normalized": false,
|
| 889 |
+
"rstrip": false,
|
| 890 |
+
"single_word": false,
|
| 891 |
+
"special": true
|
| 892 |
+
},
|
| 893 |
+
"151754": {
|
| 894 |
+
"content": "|<EXTRA_TOKENS_85>|",
|
| 895 |
+
"lstrip": false,
|
| 896 |
+
"normalized": false,
|
| 897 |
+
"rstrip": false,
|
| 898 |
+
"single_word": false,
|
| 899 |
+
"special": true
|
| 900 |
+
},
|
| 901 |
+
"151755": {
|
| 902 |
+
"content": "|<EXTRA_TOKENS_86>|",
|
| 903 |
+
"lstrip": false,
|
| 904 |
+
"normalized": false,
|
| 905 |
+
"rstrip": false,
|
| 906 |
+
"single_word": false,
|
| 907 |
+
"special": true
|
| 908 |
+
},
|
| 909 |
+
"151756": {
|
| 910 |
+
"content": "|<EXTRA_TOKENS_87>|",
|
| 911 |
+
"lstrip": false,
|
| 912 |
+
"normalized": false,
|
| 913 |
+
"rstrip": false,
|
| 914 |
+
"single_word": false,
|
| 915 |
+
"special": true
|
| 916 |
+
},
|
| 917 |
+
"151757": {
|
| 918 |
+
"content": "|<EXTRA_TOKENS_88>|",
|
| 919 |
+
"lstrip": false,
|
| 920 |
+
"normalized": false,
|
| 921 |
+
"rstrip": false,
|
| 922 |
+
"single_word": false,
|
| 923 |
+
"special": true
|
| 924 |
+
},
|
| 925 |
+
"151758": {
|
| 926 |
+
"content": "|<EXTRA_TOKENS_89>|",
|
| 927 |
+
"lstrip": false,
|
| 928 |
+
"normalized": false,
|
| 929 |
+
"rstrip": false,
|
| 930 |
+
"single_word": false,
|
| 931 |
+
"special": true
|
| 932 |
+
},
|
| 933 |
+
"151759": {
|
| 934 |
+
"content": "|<EXTRA_TOKENS_90>|",
|
| 935 |
+
"lstrip": false,
|
| 936 |
+
"normalized": false,
|
| 937 |
+
"rstrip": false,
|
| 938 |
+
"single_word": false,
|
| 939 |
+
"special": true
|
| 940 |
+
},
|
| 941 |
+
"151760": {
|
| 942 |
+
"content": "|<EXTRA_TOKENS_91>|",
|
| 943 |
+
"lstrip": false,
|
| 944 |
+
"normalized": false,
|
| 945 |
+
"rstrip": false,
|
| 946 |
+
"single_word": false,
|
| 947 |
+
"special": true
|
| 948 |
+
},
|
| 949 |
+
"151761": {
|
| 950 |
+
"content": "|<EXTRA_TOKENS_92>|",
|
| 951 |
+
"lstrip": false,
|
| 952 |
+
"normalized": false,
|
| 953 |
+
"rstrip": false,
|
| 954 |
+
"single_word": false,
|
| 955 |
+
"special": true
|
| 956 |
+
},
|
| 957 |
+
"151762": {
|
| 958 |
+
"content": "|<EXTRA_TOKENS_93>|",
|
| 959 |
+
"lstrip": false,
|
| 960 |
+
"normalized": false,
|
| 961 |
+
"rstrip": false,
|
| 962 |
+
"single_word": false,
|
| 963 |
+
"special": true
|
| 964 |
+
},
|
| 965 |
+
"151763": {
|
| 966 |
+
"content": "|<EXTRA_TOKENS_94>|",
|
| 967 |
+
"lstrip": false,
|
| 968 |
+
"normalized": false,
|
| 969 |
+
"rstrip": false,
|
| 970 |
+
"single_word": false,
|
| 971 |
+
"special": true
|
| 972 |
+
},
|
| 973 |
+
"151764": {
|
| 974 |
+
"content": "|<EXTRA_TOKENS_95>|",
|
| 975 |
+
"lstrip": false,
|
| 976 |
+
"normalized": false,
|
| 977 |
+
"rstrip": false,
|
| 978 |
+
"single_word": false,
|
| 979 |
+
"special": true
|
| 980 |
+
},
|
| 981 |
+
"151765": {
|
| 982 |
+
"content": "|<EXTRA_TOKENS_96>|",
|
| 983 |
+
"lstrip": false,
|
| 984 |
+
"normalized": false,
|
| 985 |
+
"rstrip": false,
|
| 986 |
+
"single_word": false,
|
| 987 |
+
"special": true
|
| 988 |
+
},
|
| 989 |
+
"151766": {
|
| 990 |
+
"content": "|<EXTRA_TOKENS_97>|",
|
| 991 |
+
"lstrip": false,
|
| 992 |
+
"normalized": false,
|
| 993 |
+
"rstrip": false,
|
| 994 |
+
"single_word": false,
|
| 995 |
+
"special": true
|
| 996 |
+
},
|
| 997 |
+
"151767": {
|
| 998 |
+
"content": "|<EXTRA_TOKENS_98>|",
|
| 999 |
+
"lstrip": false,
|
| 1000 |
+
"normalized": false,
|
| 1001 |
+
"rstrip": false,
|
| 1002 |
+
"single_word": false,
|
| 1003 |
+
"special": true
|
| 1004 |
+
},
|
| 1005 |
+
"151768": {
|
| 1006 |
+
"content": "|<EXTRA_TOKENS_99>|",
|
| 1007 |
+
"lstrip": false,
|
| 1008 |
+
"normalized": false,
|
| 1009 |
+
"rstrip": false,
|
| 1010 |
+
"single_word": false,
|
| 1011 |
+
"special": true
|
| 1012 |
+
},
|
| 1013 |
+
"151769": {
|
| 1014 |
+
"content": "|<EXTRA_TOKENS_100>|",
|
| 1015 |
+
"lstrip": false,
|
| 1016 |
+
"normalized": false,
|
| 1017 |
+
"rstrip": false,
|
| 1018 |
+
"single_word": false,
|
| 1019 |
+
"special": true
|
| 1020 |
+
},
|
| 1021 |
+
"151770": {
|
| 1022 |
+
"content": "|<EXTRA_TOKENS_101>|",
|
| 1023 |
+
"lstrip": false,
|
| 1024 |
+
"normalized": false,
|
| 1025 |
+
"rstrip": false,
|
| 1026 |
+
"single_word": false,
|
| 1027 |
+
"special": true
|
| 1028 |
+
},
|
| 1029 |
+
"151771": {
|
| 1030 |
+
"content": "|<EXTRA_TOKENS_102>|",
|
| 1031 |
+
"lstrip": false,
|
| 1032 |
+
"normalized": false,
|
| 1033 |
+
"rstrip": false,
|
| 1034 |
+
"single_word": false,
|
| 1035 |
+
"special": true
|
| 1036 |
+
},
|
| 1037 |
+
"151772": {
|
| 1038 |
+
"content": "|<EXTRA_TOKENS_103>|",
|
| 1039 |
+
"lstrip": false,
|
| 1040 |
+
"normalized": false,
|
| 1041 |
+
"rstrip": false,
|
| 1042 |
+
"single_word": false,
|
| 1043 |
+
"special": true
|
| 1044 |
+
},
|
| 1045 |
+
"151773": {
|
| 1046 |
+
"content": "|<EXTRA_TOKENS_104>|",
|
| 1047 |
+
"lstrip": false,
|
| 1048 |
+
"normalized": false,
|
| 1049 |
+
"rstrip": false,
|
| 1050 |
+
"single_word": false,
|
| 1051 |
+
"special": true
|
| 1052 |
+
},
|
| 1053 |
+
"151774": {
|
| 1054 |
+
"content": "|<EXTRA_TOKENS_105>|",
|
| 1055 |
+
"lstrip": false,
|
| 1056 |
+
"normalized": false,
|
| 1057 |
+
"rstrip": false,
|
| 1058 |
+
"single_word": false,
|
| 1059 |
+
"special": true
|
| 1060 |
+
},
|
| 1061 |
+
"151775": {
|
| 1062 |
+
"content": "|<EXTRA_TOKENS_106>|",
|
| 1063 |
+
"lstrip": false,
|
| 1064 |
+
"normalized": false,
|
| 1065 |
+
"rstrip": false,
|
| 1066 |
+
"single_word": false,
|
| 1067 |
+
"special": true
|
| 1068 |
+
},
|
| 1069 |
+
"151776": {
|
| 1070 |
+
"content": "|<EXTRA_TOKENS_107>|",
|
| 1071 |
+
"lstrip": false,
|
| 1072 |
+
"normalized": false,
|
| 1073 |
+
"rstrip": false,
|
| 1074 |
+
"single_word": false,
|
| 1075 |
+
"special": true
|
| 1076 |
+
},
|
| 1077 |
+
"151777": {
|
| 1078 |
+
"content": "|<EXTRA_TOKENS_108>|",
|
| 1079 |
+
"lstrip": false,
|
| 1080 |
+
"normalized": false,
|
| 1081 |
+
"rstrip": false,
|
| 1082 |
+
"single_word": false,
|
| 1083 |
+
"special": true
|
| 1084 |
+
},
|
| 1085 |
+
"151778": {
|
| 1086 |
+
"content": "|<EXTRA_TOKENS_109>|",
|
| 1087 |
+
"lstrip": false,
|
| 1088 |
+
"normalized": false,
|
| 1089 |
+
"rstrip": false,
|
| 1090 |
+
"single_word": false,
|
| 1091 |
+
"special": true
|
| 1092 |
+
},
|
| 1093 |
+
"151779": {
|
| 1094 |
+
"content": "|<EXTRA_TOKENS_110>|",
|
| 1095 |
+
"lstrip": false,
|
| 1096 |
+
"normalized": false,
|
| 1097 |
+
"rstrip": false,
|
| 1098 |
+
"single_word": false,
|
| 1099 |
+
"special": true
|
| 1100 |
+
},
|
| 1101 |
+
"151780": {
|
| 1102 |
+
"content": "|<EXTRA_TOKENS_111>|",
|
| 1103 |
+
"lstrip": false,
|
| 1104 |
+
"normalized": false,
|
| 1105 |
+
"rstrip": false,
|
| 1106 |
+
"single_word": false,
|
| 1107 |
+
"special": true
|
| 1108 |
+
},
|
| 1109 |
+
"151781": {
|
| 1110 |
+
"content": "|<EXTRA_TOKENS_112>|",
|
| 1111 |
+
"lstrip": false,
|
| 1112 |
+
"normalized": false,
|
| 1113 |
+
"rstrip": false,
|
| 1114 |
+
"single_word": false,
|
| 1115 |
+
"special": true
|
| 1116 |
+
},
|
| 1117 |
+
"151782": {
|
| 1118 |
+
"content": "|<EXTRA_TOKENS_113>|",
|
| 1119 |
+
"lstrip": false,
|
| 1120 |
+
"normalized": false,
|
| 1121 |
+
"rstrip": false,
|
| 1122 |
+
"single_word": false,
|
| 1123 |
+
"special": true
|
| 1124 |
+
},
|
| 1125 |
+
"151783": {
|
| 1126 |
+
"content": "|<EXTRA_TOKENS_114>|",
|
| 1127 |
+
"lstrip": false,
|
| 1128 |
+
"normalized": false,
|
| 1129 |
+
"rstrip": false,
|
| 1130 |
+
"single_word": false,
|
| 1131 |
+
"special": true
|
| 1132 |
+
},
|
| 1133 |
+
"151784": {
|
| 1134 |
+
"content": "|<EXTRA_TOKENS_115>|",
|
| 1135 |
+
"lstrip": false,
|
| 1136 |
+
"normalized": false,
|
| 1137 |
+
"rstrip": false,
|
| 1138 |
+
"single_word": false,
|
| 1139 |
+
"special": true
|
| 1140 |
+
},
|
| 1141 |
+
"151785": {
|
| 1142 |
+
"content": "|<EXTRA_TOKENS_116>|",
|
| 1143 |
+
"lstrip": false,
|
| 1144 |
+
"normalized": false,
|
| 1145 |
+
"rstrip": false,
|
| 1146 |
+
"single_word": false,
|
| 1147 |
+
"special": true
|
| 1148 |
+
},
|
| 1149 |
+
"151786": {
|
| 1150 |
+
"content": "|<EXTRA_TOKENS_117>|",
|
| 1151 |
+
"lstrip": false,
|
| 1152 |
+
"normalized": false,
|
| 1153 |
+
"rstrip": false,
|
| 1154 |
+
"single_word": false,
|
| 1155 |
+
"special": true
|
| 1156 |
+
},
|
| 1157 |
+
"151787": {
|
| 1158 |
+
"content": "|<EXTRA_TOKENS_118>|",
|
| 1159 |
+
"lstrip": false,
|
| 1160 |
+
"normalized": false,
|
| 1161 |
+
"rstrip": false,
|
| 1162 |
+
"single_word": false,
|
| 1163 |
+
"special": true
|
| 1164 |
+
},
|
| 1165 |
+
"151788": {
|
| 1166 |
+
"content": "|<EXTRA_TOKENS_119>|",
|
| 1167 |
+
"lstrip": false,
|
| 1168 |
+
"normalized": false,
|
| 1169 |
+
"rstrip": false,
|
| 1170 |
+
"single_word": false,
|
| 1171 |
+
"special": true
|
| 1172 |
+
},
|
| 1173 |
+
"151789": {
|
| 1174 |
+
"content": "|<EXTRA_TOKENS_120>|",
|
| 1175 |
+
"lstrip": false,
|
| 1176 |
+
"normalized": false,
|
| 1177 |
+
"rstrip": false,
|
| 1178 |
+
"single_word": false,
|
| 1179 |
+
"special": true
|
| 1180 |
+
},
|
| 1181 |
+
"151790": {
|
| 1182 |
+
"content": "|<EXTRA_TOKENS_121>|",
|
| 1183 |
+
"lstrip": false,
|
| 1184 |
+
"normalized": false,
|
| 1185 |
+
"rstrip": false,
|
| 1186 |
+
"single_word": false,
|
| 1187 |
+
"special": true
|
| 1188 |
+
},
|
| 1189 |
+
"151791": {
|
| 1190 |
+
"content": "|<EXTRA_TOKENS_122>|",
|
| 1191 |
+
"lstrip": false,
|
| 1192 |
+
"normalized": false,
|
| 1193 |
+
"rstrip": false,
|
| 1194 |
+
"single_word": false,
|
| 1195 |
+
"special": true
|
| 1196 |
+
},
|
| 1197 |
+
"151792": {
|
| 1198 |
+
"content": "|<EXTRA_TOKENS_123>|",
|
| 1199 |
+
"lstrip": false,
|
| 1200 |
+
"normalized": false,
|
| 1201 |
+
"rstrip": false,
|
| 1202 |
+
"single_word": false,
|
| 1203 |
+
"special": true
|
| 1204 |
+
},
|
| 1205 |
+
"151793": {
|
| 1206 |
+
"content": "|<EXTRA_TOKENS_124>|",
|
| 1207 |
+
"lstrip": false,
|
| 1208 |
+
"normalized": false,
|
| 1209 |
+
"rstrip": false,
|
| 1210 |
+
"single_word": false,
|
| 1211 |
+
"special": true
|
| 1212 |
+
},
|
| 1213 |
+
"151794": {
|
| 1214 |
+
"content": "|<EXTRA_TOKENS_125>|",
|
| 1215 |
+
"lstrip": false,
|
| 1216 |
+
"normalized": false,
|
| 1217 |
+
"rstrip": false,
|
| 1218 |
+
"single_word": false,
|
| 1219 |
+
"special": true
|
| 1220 |
+
},
|
| 1221 |
+
"151795": {
|
| 1222 |
+
"content": "|<EXTRA_TOKENS_126>|",
|
| 1223 |
+
"lstrip": false,
|
| 1224 |
+
"normalized": false,
|
| 1225 |
+
"rstrip": false,
|
| 1226 |
+
"single_word": false,
|
| 1227 |
+
"special": true
|
| 1228 |
+
},
|
| 1229 |
+
"151796": {
|
| 1230 |
+
"content": "|<EXTRA_TOKENS_127>|",
|
| 1231 |
+
"lstrip": false,
|
| 1232 |
+
"normalized": false,
|
| 1233 |
+
"rstrip": false,
|
| 1234 |
+
"single_word": false,
|
| 1235 |
+
"special": true
|
| 1236 |
+
},
|
| 1237 |
+
"151797": {
|
| 1238 |
+
"content": "|<EXTRA_TOKENS_128>|",
|
| 1239 |
+
"lstrip": false,
|
| 1240 |
+
"normalized": false,
|
| 1241 |
+
"rstrip": false,
|
| 1242 |
+
"single_word": false,
|
| 1243 |
+
"special": true
|
| 1244 |
+
},
|
| 1245 |
+
"151798": {
|
| 1246 |
+
"content": "|<EXTRA_TOKENS_129>|",
|
| 1247 |
+
"lstrip": false,
|
| 1248 |
+
"normalized": false,
|
| 1249 |
+
"rstrip": false,
|
| 1250 |
+
"single_word": false,
|
| 1251 |
+
"special": true
|
| 1252 |
+
},
|
| 1253 |
+
"151799": {
|
| 1254 |
+
"content": "|<EXTRA_TOKENS_130>|",
|
| 1255 |
+
"lstrip": false,
|
| 1256 |
+
"normalized": false,
|
| 1257 |
+
"rstrip": false,
|
| 1258 |
+
"single_word": false,
|
| 1259 |
+
"special": true
|
| 1260 |
+
},
|
| 1261 |
+
"151800": {
|
| 1262 |
+
"content": "|<EXTRA_TOKENS_131>|",
|
| 1263 |
+
"lstrip": false,
|
| 1264 |
+
"normalized": false,
|
| 1265 |
+
"rstrip": false,
|
| 1266 |
+
"single_word": false,
|
| 1267 |
+
"special": true
|
| 1268 |
+
},
|
| 1269 |
+
"151801": {
|
| 1270 |
+
"content": "|<EXTRA_TOKENS_132>|",
|
| 1271 |
+
"lstrip": false,
|
| 1272 |
+
"normalized": false,
|
| 1273 |
+
"rstrip": false,
|
| 1274 |
+
"single_word": false,
|
| 1275 |
+
"special": true
|
| 1276 |
+
},
|
| 1277 |
+
"151802": {
|
| 1278 |
+
"content": "|<EXTRA_TOKENS_133>|",
|
| 1279 |
+
"lstrip": false,
|
| 1280 |
+
"normalized": false,
|
| 1281 |
+
"rstrip": false,
|
| 1282 |
+
"single_word": false,
|
| 1283 |
+
"special": true
|
| 1284 |
+
},
|
| 1285 |
+
"151803": {
|
| 1286 |
+
"content": "|<EXTRA_TOKENS_134>|",
|
| 1287 |
+
"lstrip": false,
|
| 1288 |
+
"normalized": false,
|
| 1289 |
+
"rstrip": false,
|
| 1290 |
+
"single_word": false,
|
| 1291 |
+
"special": true
|
| 1292 |
+
},
|
| 1293 |
+
"151804": {
|
| 1294 |
+
"content": "|<EXTRA_TOKENS_135>|",
|
| 1295 |
+
"lstrip": false,
|
| 1296 |
+
"normalized": false,
|
| 1297 |
+
"rstrip": false,
|
| 1298 |
+
"single_word": false,
|
| 1299 |
+
"special": true
|
| 1300 |
+
},
|
| 1301 |
+
"151805": {
|
| 1302 |
+
"content": "|<EXTRA_TOKENS_136>|",
|
| 1303 |
+
"lstrip": false,
|
| 1304 |
+
"normalized": false,
|
| 1305 |
+
"rstrip": false,
|
| 1306 |
+
"single_word": false,
|
| 1307 |
+
"special": true
|
| 1308 |
+
},
|
| 1309 |
+
"151806": {
|
| 1310 |
+
"content": "|<EXTRA_TOKENS_137>|",
|
| 1311 |
+
"lstrip": false,
|
| 1312 |
+
"normalized": false,
|
| 1313 |
+
"rstrip": false,
|
| 1314 |
+
"single_word": false,
|
| 1315 |
+
"special": true
|
| 1316 |
+
},
|
| 1317 |
+
"151807": {
|
| 1318 |
+
"content": "|<EXTRA_TOKENS_138>|",
|
| 1319 |
+
"lstrip": false,
|
| 1320 |
+
"normalized": false,
|
| 1321 |
+
"rstrip": false,
|
| 1322 |
+
"single_word": false,
|
| 1323 |
+
"special": true
|
| 1324 |
+
},
|
| 1325 |
+
"151808": {
|
| 1326 |
+
"content": "|<EXTRA_TOKENS_139>|",
|
| 1327 |
+
"lstrip": false,
|
| 1328 |
+
"normalized": false,
|
| 1329 |
+
"rstrip": false,
|
| 1330 |
+
"single_word": false,
|
| 1331 |
+
"special": true
|
| 1332 |
+
},
|
| 1333 |
+
"151809": {
|
| 1334 |
+
"content": "|<EXTRA_TOKENS_140>|",
|
| 1335 |
+
"lstrip": false,
|
| 1336 |
+
"normalized": false,
|
| 1337 |
+
"rstrip": false,
|
| 1338 |
+
"single_word": false,
|
| 1339 |
+
"special": true
|
| 1340 |
+
},
|
| 1341 |
+
"151810": {
|
| 1342 |
+
"content": "|<EXTRA_TOKENS_141>|",
|
| 1343 |
+
"lstrip": false,
|
| 1344 |
+
"normalized": false,
|
| 1345 |
+
"rstrip": false,
|
| 1346 |
+
"single_word": false,
|
| 1347 |
+
"special": true
|
| 1348 |
+
},
|
| 1349 |
+
"151811": {
|
| 1350 |
+
"content": "|<EXTRA_TOKENS_142>|",
|
| 1351 |
+
"lstrip": false,
|
| 1352 |
+
"normalized": false,
|
| 1353 |
+
"rstrip": false,
|
| 1354 |
+
"single_word": false,
|
| 1355 |
+
"special": true
|
| 1356 |
+
},
|
| 1357 |
+
"151812": {
|
| 1358 |
+
"content": "|<EXTRA_TOKENS_143>|",
|
| 1359 |
+
"lstrip": false,
|
| 1360 |
+
"normalized": false,
|
| 1361 |
+
"rstrip": false,
|
| 1362 |
+
"single_word": false,
|
| 1363 |
+
"special": true
|
| 1364 |
+
},
|
| 1365 |
+
"151813": {
|
| 1366 |
+
"content": "|<EXTRA_TOKENS_144>|",
|
| 1367 |
+
"lstrip": false,
|
| 1368 |
+
"normalized": false,
|
| 1369 |
+
"rstrip": false,
|
| 1370 |
+
"single_word": false,
|
| 1371 |
+
"special": true
|
| 1372 |
+
},
|
| 1373 |
+
"151814": {
|
| 1374 |
+
"content": "|<EXTRA_TOKENS_145>|",
|
| 1375 |
+
"lstrip": false,
|
| 1376 |
+
"normalized": false,
|
| 1377 |
+
"rstrip": false,
|
| 1378 |
+
"single_word": false,
|
| 1379 |
+
"special": true
|
| 1380 |
+
},
|
| 1381 |
+
"151815": {
|
| 1382 |
+
"content": "|<EXTRA_TOKENS_146>|",
|
| 1383 |
+
"lstrip": false,
|
| 1384 |
+
"normalized": false,
|
| 1385 |
+
"rstrip": false,
|
| 1386 |
+
"single_word": false,
|
| 1387 |
+
"special": true
|
| 1388 |
+
},
|
| 1389 |
+
"151816": {
|
| 1390 |
+
"content": "|<EXTRA_TOKENS_147>|",
|
| 1391 |
+
"lstrip": false,
|
| 1392 |
+
"normalized": false,
|
| 1393 |
+
"rstrip": false,
|
| 1394 |
+
"single_word": false,
|
| 1395 |
+
"special": true
|
| 1396 |
+
},
|
| 1397 |
+
"151817": {
|
| 1398 |
+
"content": "|<EXTRA_TOKENS_148>|",
|
| 1399 |
+
"lstrip": false,
|
| 1400 |
+
"normalized": false,
|
| 1401 |
+
"rstrip": false,
|
| 1402 |
+
"single_word": false,
|
| 1403 |
+
"special": true
|
| 1404 |
+
},
|
| 1405 |
+
"151818": {
|
| 1406 |
+
"content": "|<EXTRA_TOKENS_149>|",
|
| 1407 |
+
"lstrip": false,
|
| 1408 |
+
"normalized": false,
|
| 1409 |
+
"rstrip": false,
|
| 1410 |
+
"single_word": false,
|
| 1411 |
+
"special": true
|
| 1412 |
+
},
|
| 1413 |
+
"151819": {
|
| 1414 |
+
"content": "|<EXTRA_TOKENS_150>|",
|
| 1415 |
+
"lstrip": false,
|
| 1416 |
+
"normalized": false,
|
| 1417 |
+
"rstrip": false,
|
| 1418 |
+
"single_word": false,
|
| 1419 |
+
"special": true
|
| 1420 |
+
},
|
| 1421 |
+
"151820": {
|
| 1422 |
+
"content": "|<EXTRA_TOKENS_151>|",
|
| 1423 |
+
"lstrip": false,
|
| 1424 |
+
"normalized": false,
|
| 1425 |
+
"rstrip": false,
|
| 1426 |
+
"single_word": false,
|
| 1427 |
+
"special": true
|
| 1428 |
+
},
|
| 1429 |
+
"151821": {
|
| 1430 |
+
"content": "|<EXTRA_TOKENS_152>|",
|
| 1431 |
+
"lstrip": false,
|
| 1432 |
+
"normalized": false,
|
| 1433 |
+
"rstrip": false,
|
| 1434 |
+
"single_word": false,
|
| 1435 |
+
"special": true
|
| 1436 |
+
},
|
| 1437 |
+
"151822": {
|
| 1438 |
+
"content": "|<EXTRA_TOKENS_153>|",
|
| 1439 |
+
"lstrip": false,
|
| 1440 |
+
"normalized": false,
|
| 1441 |
+
"rstrip": false,
|
| 1442 |
+
"single_word": false,
|
| 1443 |
+
"special": true
|
| 1444 |
+
},
|
| 1445 |
+
"151823": {
|
| 1446 |
+
"content": "|<EXTRA_TOKENS_154>|",
|
| 1447 |
+
"lstrip": false,
|
| 1448 |
+
"normalized": false,
|
| 1449 |
+
"rstrip": false,
|
| 1450 |
+
"single_word": false,
|
| 1451 |
+
"special": true
|
| 1452 |
+
},
|
| 1453 |
+
"151824": {
|
| 1454 |
+
"content": "|<EXTRA_TOKENS_155>|",
|
| 1455 |
+
"lstrip": false,
|
| 1456 |
+
"normalized": false,
|
| 1457 |
+
"rstrip": false,
|
| 1458 |
+
"single_word": false,
|
| 1459 |
+
"special": true
|
| 1460 |
+
},
|
| 1461 |
+
"151825": {
|
| 1462 |
+
"content": "|<EXTRA_TOKENS_156>|",
|
| 1463 |
+
"lstrip": false,
|
| 1464 |
+
"normalized": false,
|
| 1465 |
+
"rstrip": false,
|
| 1466 |
+
"single_word": false,
|
| 1467 |
+
"special": true
|
| 1468 |
+
},
|
| 1469 |
+
"151826": {
|
| 1470 |
+
"content": "|<EXTRA_TOKENS_157>|",
|
| 1471 |
+
"lstrip": false,
|
| 1472 |
+
"normalized": false,
|
| 1473 |
+
"rstrip": false,
|
| 1474 |
+
"single_word": false,
|
| 1475 |
+
"special": true
|
| 1476 |
+
},
|
| 1477 |
+
"151827": {
|
| 1478 |
+
"content": "|<EXTRA_TOKENS_158>|",
|
| 1479 |
+
"lstrip": false,
|
| 1480 |
+
"normalized": false,
|
| 1481 |
+
"rstrip": false,
|
| 1482 |
+
"single_word": false,
|
| 1483 |
+
"special": true
|
| 1484 |
+
},
|
| 1485 |
+
"151828": {
|
| 1486 |
+
"content": "|<EXTRA_TOKENS_159>|",
|
| 1487 |
+
"lstrip": false,
|
| 1488 |
+
"normalized": false,
|
| 1489 |
+
"rstrip": false,
|
| 1490 |
+
"single_word": false,
|
| 1491 |
+
"special": true
|
| 1492 |
+
},
|
| 1493 |
+
"151829": {
|
| 1494 |
+
"content": "|<EXTRA_TOKENS_160>|",
|
| 1495 |
+
"lstrip": false,
|
| 1496 |
+
"normalized": false,
|
| 1497 |
+
"rstrip": false,
|
| 1498 |
+
"single_word": false,
|
| 1499 |
+
"special": true
|
| 1500 |
+
},
|
| 1501 |
+
"151830": {
|
| 1502 |
+
"content": "|<EXTRA_TOKENS_161>|",
|
| 1503 |
+
"lstrip": false,
|
| 1504 |
+
"normalized": false,
|
| 1505 |
+
"rstrip": false,
|
| 1506 |
+
"single_word": false,
|
| 1507 |
+
"special": true
|
| 1508 |
+
},
|
| 1509 |
+
"151831": {
|
| 1510 |
+
"content": "|<EXTRA_TOKENS_162>|",
|
| 1511 |
+
"lstrip": false,
|
| 1512 |
+
"normalized": false,
|
| 1513 |
+
"rstrip": false,
|
| 1514 |
+
"single_word": false,
|
| 1515 |
+
"special": true
|
| 1516 |
+
},
|
| 1517 |
+
"151832": {
|
| 1518 |
+
"content": "|<EXTRA_TOKENS_163>|",
|
| 1519 |
+
"lstrip": false,
|
| 1520 |
+
"normalized": false,
|
| 1521 |
+
"rstrip": false,
|
| 1522 |
+
"single_word": false,
|
| 1523 |
+
"special": true
|
| 1524 |
+
},
|
| 1525 |
+
"151833": {
|
| 1526 |
+
"content": "|<EXTRA_TOKENS_164>|",
|
| 1527 |
+
"lstrip": false,
|
| 1528 |
+
"normalized": false,
|
| 1529 |
+
"rstrip": false,
|
| 1530 |
+
"single_word": false,
|
| 1531 |
+
"special": true
|
| 1532 |
+
},
|
| 1533 |
+
"151834": {
|
| 1534 |
+
"content": "|<EXTRA_TOKENS_165>|",
|
| 1535 |
+
"lstrip": false,
|
| 1536 |
+
"normalized": false,
|
| 1537 |
+
"rstrip": false,
|
| 1538 |
+
"single_word": false,
|
| 1539 |
+
"special": true
|
| 1540 |
+
},
|
| 1541 |
+
"151835": {
|
| 1542 |
+
"content": "|<EXTRA_TOKENS_166>|",
|
| 1543 |
+
"lstrip": false,
|
| 1544 |
+
"normalized": false,
|
| 1545 |
+
"rstrip": false,
|
| 1546 |
+
"single_word": false,
|
| 1547 |
+
"special": true
|
| 1548 |
+
},
|
| 1549 |
+
"151836": {
|
| 1550 |
+
"content": "|<EXTRA_TOKENS_167>|",
|
| 1551 |
+
"lstrip": false,
|
| 1552 |
+
"normalized": false,
|
| 1553 |
+
"rstrip": false,
|
| 1554 |
+
"single_word": false,
|
| 1555 |
+
"special": true
|
| 1556 |
+
},
|
| 1557 |
+
"151837": {
|
| 1558 |
+
"content": "|<EXTRA_TOKENS_168>|",
|
| 1559 |
+
"lstrip": false,
|
| 1560 |
+
"normalized": false,
|
| 1561 |
+
"rstrip": false,
|
| 1562 |
+
"single_word": false,
|
| 1563 |
+
"special": true
|
| 1564 |
+
},
|
| 1565 |
+
"151838": {
|
| 1566 |
+
"content": "|<EXTRA_TOKENS_169>|",
|
| 1567 |
+
"lstrip": false,
|
| 1568 |
+
"normalized": false,
|
| 1569 |
+
"rstrip": false,
|
| 1570 |
+
"single_word": false,
|
| 1571 |
+
"special": true
|
| 1572 |
+
},
|
| 1573 |
+
"151839": {
|
| 1574 |
+
"content": "|<EXTRA_TOKENS_170>|",
|
| 1575 |
+
"lstrip": false,
|
| 1576 |
+
"normalized": false,
|
| 1577 |
+
"rstrip": false,
|
| 1578 |
+
"single_word": false,
|
| 1579 |
+
"special": true
|
| 1580 |
+
},
|
| 1581 |
+
"151840": {
|
| 1582 |
+
"content": "|<EXTRA_TOKENS_171>|",
|
| 1583 |
+
"lstrip": false,
|
| 1584 |
+
"normalized": false,
|
| 1585 |
+
"rstrip": false,
|
| 1586 |
+
"single_word": false,
|
| 1587 |
+
"special": true
|
| 1588 |
+
},
|
| 1589 |
+
"151841": {
|
| 1590 |
+
"content": "|<EXTRA_TOKENS_172>|",
|
| 1591 |
+
"lstrip": false,
|
| 1592 |
+
"normalized": false,
|
| 1593 |
+
"rstrip": false,
|
| 1594 |
+
"single_word": false,
|
| 1595 |
+
"special": true
|
| 1596 |
+
},
|
| 1597 |
+
"151842": {
|
| 1598 |
+
"content": "|<EXTRA_TOKENS_173>|",
|
| 1599 |
+
"lstrip": false,
|
| 1600 |
+
"normalized": false,
|
| 1601 |
+
"rstrip": false,
|
| 1602 |
+
"single_word": false,
|
| 1603 |
+
"special": true
|
| 1604 |
+
},
|
| 1605 |
+
"151843": {
|
| 1606 |
+
"content": "|<EXTRA_TOKENS_174>|",
|
| 1607 |
+
"lstrip": false,
|
| 1608 |
+
"normalized": false,
|
| 1609 |
+
"rstrip": false,
|
| 1610 |
+
"single_word": false,
|
| 1611 |
+
"special": true
|
| 1612 |
+
},
|
| 1613 |
+
"151844": {
|
| 1614 |
+
"content": "|<EXTRA_TOKENS_175>|",
|
| 1615 |
+
"lstrip": false,
|
| 1616 |
+
"normalized": false,
|
| 1617 |
+
"rstrip": false,
|
| 1618 |
+
"single_word": false,
|
| 1619 |
+
"special": true
|
| 1620 |
+
},
|
| 1621 |
+
"151845": {
|
| 1622 |
+
"content": "|<EXTRA_TOKENS_176>|",
|
| 1623 |
+
"lstrip": false,
|
| 1624 |
+
"normalized": false,
|
| 1625 |
+
"rstrip": false,
|
| 1626 |
+
"single_word": false,
|
| 1627 |
+
"special": true
|
| 1628 |
+
},
|
| 1629 |
+
"151846": {
|
| 1630 |
+
"content": "|<EXTRA_TOKENS_177>|",
|
| 1631 |
+
"lstrip": false,
|
| 1632 |
+
"normalized": false,
|
| 1633 |
+
"rstrip": false,
|
| 1634 |
+
"single_word": false,
|
| 1635 |
+
"special": true
|
| 1636 |
+
},
|
| 1637 |
+
"151847": {
|
| 1638 |
+
"content": "|<EXTRA_TOKENS_178>|",
|
| 1639 |
+
"lstrip": false,
|
| 1640 |
+
"normalized": false,
|
| 1641 |
+
"rstrip": false,
|
| 1642 |
+
"single_word": false,
|
| 1643 |
+
"special": true
|
| 1644 |
+
},
|
| 1645 |
+
"151848": {
|
| 1646 |
+
"content": "|<EXTRA_TOKENS_179>|",
|
| 1647 |
+
"lstrip": false,
|
| 1648 |
+
"normalized": false,
|
| 1649 |
+
"rstrip": false,
|
| 1650 |
+
"single_word": false,
|
| 1651 |
+
"special": true
|
| 1652 |
+
},
|
| 1653 |
+
"151849": {
|
| 1654 |
+
"content": "|<EXTRA_TOKENS_180>|",
|
| 1655 |
+
"lstrip": false,
|
| 1656 |
+
"normalized": false,
|
| 1657 |
+
"rstrip": false,
|
| 1658 |
+
"single_word": false,
|
| 1659 |
+
"special": true
|
| 1660 |
+
},
|
| 1661 |
+
"151850": {
|
| 1662 |
+
"content": "|<EXTRA_TOKENS_181>|",
|
| 1663 |
+
"lstrip": false,
|
| 1664 |
+
"normalized": false,
|
| 1665 |
+
"rstrip": false,
|
| 1666 |
+
"single_word": false,
|
| 1667 |
+
"special": true
|
| 1668 |
+
},
|
| 1669 |
+
"151851": {
|
| 1670 |
+
"content": "|<EXTRA_TOKENS_182>|",
|
| 1671 |
+
"lstrip": false,
|
| 1672 |
+
"normalized": false,
|
| 1673 |
+
"rstrip": false,
|
| 1674 |
+
"single_word": false,
|
| 1675 |
+
"special": true
|
| 1676 |
+
},
|
| 1677 |
+
"151852": {
|
| 1678 |
+
"content": "|<EXTRA_TOKENS_183>|",
|
| 1679 |
+
"lstrip": false,
|
| 1680 |
+
"normalized": false,
|
| 1681 |
+
"rstrip": false,
|
| 1682 |
+
"single_word": false,
|
| 1683 |
+
"special": true
|
| 1684 |
+
},
|
| 1685 |
+
"151853": {
|
| 1686 |
+
"content": "|<EXTRA_TOKENS_184>|",
|
| 1687 |
+
"lstrip": false,
|
| 1688 |
+
"normalized": false,
|
| 1689 |
+
"rstrip": false,
|
| 1690 |
+
"single_word": false,
|
| 1691 |
+
"special": true
|
| 1692 |
+
},
|
| 1693 |
+
"151854": {
|
| 1694 |
+
"content": "|<EXTRA_TOKENS_185>|",
|
| 1695 |
+
"lstrip": false,
|
| 1696 |
+
"normalized": false,
|
| 1697 |
+
"rstrip": false,
|
| 1698 |
+
"single_word": false,
|
| 1699 |
+
"special": true
|
| 1700 |
+
},
|
| 1701 |
+
"151855": {
|
| 1702 |
+
"content": "|<EXTRA_TOKENS_186>|",
|
| 1703 |
+
"lstrip": false,
|
| 1704 |
+
"normalized": false,
|
| 1705 |
+
"rstrip": false,
|
| 1706 |
+
"single_word": false,
|
| 1707 |
+
"special": true
|
| 1708 |
+
},
|
| 1709 |
+
"151856": {
|
| 1710 |
+
"content": "|<EXTRA_TOKENS_187>|",
|
| 1711 |
+
"lstrip": false,
|
| 1712 |
+
"normalized": false,
|
| 1713 |
+
"rstrip": false,
|
| 1714 |
+
"single_word": false,
|
| 1715 |
+
"special": true
|
| 1716 |
+
},
|
| 1717 |
+
"151857": {
|
| 1718 |
+
"content": "|<EXTRA_TOKENS_188>|",
|
| 1719 |
+
"lstrip": false,
|
| 1720 |
+
"normalized": false,
|
| 1721 |
+
"rstrip": false,
|
| 1722 |
+
"single_word": false,
|
| 1723 |
+
"special": true
|
| 1724 |
+
},
|
| 1725 |
+
"151858": {
|
| 1726 |
+
"content": "|<EXTRA_TOKENS_189>|",
|
| 1727 |
+
"lstrip": false,
|
| 1728 |
+
"normalized": false,
|
| 1729 |
+
"rstrip": false,
|
| 1730 |
+
"single_word": false,
|
| 1731 |
+
"special": true
|
| 1732 |
+
},
|
| 1733 |
+
"151859": {
|
| 1734 |
+
"content": "|<EXTRA_TOKENS_190>|",
|
| 1735 |
+
"lstrip": false,
|
| 1736 |
+
"normalized": false,
|
| 1737 |
+
"rstrip": false,
|
| 1738 |
+
"single_word": false,
|
| 1739 |
+
"special": true
|
| 1740 |
+
},
|
| 1741 |
+
"151860": {
|
| 1742 |
+
"content": "|<EXTRA_TOKENS_191>|",
|
| 1743 |
+
"lstrip": false,
|
| 1744 |
+
"normalized": false,
|
| 1745 |
+
"rstrip": false,
|
| 1746 |
+
"single_word": false,
|
| 1747 |
+
"special": true
|
| 1748 |
+
},
|
| 1749 |
+
"151861": {
|
| 1750 |
+
"content": "|<EXTRA_TOKENS_192>|",
|
| 1751 |
+
"lstrip": false,
|
| 1752 |
+
"normalized": false,
|
| 1753 |
+
"rstrip": false,
|
| 1754 |
+
"single_word": false,
|
| 1755 |
+
"special": true
|
| 1756 |
+
},
|
| 1757 |
+
"151862": {
|
| 1758 |
+
"content": "|<EXTRA_TOKENS_193>|",
|
| 1759 |
+
"lstrip": false,
|
| 1760 |
+
"normalized": false,
|
| 1761 |
+
"rstrip": false,
|
| 1762 |
+
"single_word": false,
|
| 1763 |
+
"special": true
|
| 1764 |
+
},
|
| 1765 |
+
"151863": {
|
| 1766 |
+
"content": "|<EXTRA_TOKENS_194>|",
|
| 1767 |
+
"lstrip": false,
|
| 1768 |
+
"normalized": false,
|
| 1769 |
+
"rstrip": false,
|
| 1770 |
+
"single_word": false,
|
| 1771 |
+
"special": true
|
| 1772 |
+
},
|
| 1773 |
+
"151864": {
|
| 1774 |
+
"content": "|<EXTRA_TOKENS_195>|",
|
| 1775 |
+
"lstrip": false,
|
| 1776 |
+
"normalized": false,
|
| 1777 |
+
"rstrip": false,
|
| 1778 |
+
"single_word": false,
|
| 1779 |
+
"special": true
|
| 1780 |
+
},
|
| 1781 |
+
"151865": {
|
| 1782 |
+
"content": "|<EXTRA_TOKENS_196>|",
|
| 1783 |
+
"lstrip": false,
|
| 1784 |
+
"normalized": false,
|
| 1785 |
+
"rstrip": false,
|
| 1786 |
+
"single_word": false,
|
| 1787 |
+
"special": true
|
| 1788 |
+
},
|
| 1789 |
+
"151866": {
|
| 1790 |
+
"content": "|<EXTRA_TOKENS_197>|",
|
| 1791 |
+
"lstrip": false,
|
| 1792 |
+
"normalized": false,
|
| 1793 |
+
"rstrip": false,
|
| 1794 |
+
"single_word": false,
|
| 1795 |
+
"special": true
|
| 1796 |
+
},
|
| 1797 |
+
"151867": {
|
| 1798 |
+
"content": "|<EXTRA_TOKENS_198>|",
|
| 1799 |
+
"lstrip": false,
|
| 1800 |
+
"normalized": false,
|
| 1801 |
+
"rstrip": false,
|
| 1802 |
+
"single_word": false,
|
| 1803 |
+
"special": true
|
| 1804 |
+
},
|
| 1805 |
+
"151868": {
|
| 1806 |
+
"content": "|<EXTRA_TOKENS_199>|",
|
| 1807 |
+
"lstrip": false,
|
| 1808 |
+
"normalized": false,
|
| 1809 |
+
"rstrip": false,
|
| 1810 |
+
"single_word": false,
|
| 1811 |
+
"special": true
|
| 1812 |
+
},
|
| 1813 |
+
"151869": {
|
| 1814 |
+
"content": "|<EXTRA_TOKENS_200>|",
|
| 1815 |
+
"lstrip": false,
|
| 1816 |
+
"normalized": false,
|
| 1817 |
+
"rstrip": false,
|
| 1818 |
+
"single_word": false,
|
| 1819 |
+
"special": true
|
| 1820 |
+
},
|
| 1821 |
+
"151870": {
|
| 1822 |
+
"content": "|<EXTRA_TOKENS_201>|",
|
| 1823 |
+
"lstrip": false,
|
| 1824 |
+
"normalized": false,
|
| 1825 |
+
"rstrip": false,
|
| 1826 |
+
"single_word": false,
|
| 1827 |
+
"special": true
|
| 1828 |
+
},
|
| 1829 |
+
"151871": {
|
| 1830 |
+
"content": "|<EXTRA_TOKENS_202>|",
|
| 1831 |
+
"lstrip": false,
|
| 1832 |
+
"normalized": false,
|
| 1833 |
+
"rstrip": false,
|
| 1834 |
+
"single_word": false,
|
| 1835 |
+
"special": true
|
| 1836 |
+
},
|
| 1837 |
+
"151872": {
|
| 1838 |
+
"content": "|<EXTRA_TOKENS_203>|",
|
| 1839 |
+
"lstrip": false,
|
| 1840 |
+
"normalized": false,
|
| 1841 |
+
"rstrip": false,
|
| 1842 |
+
"single_word": false,
|
| 1843 |
+
"special": true
|
| 1844 |
+
},
|
| 1845 |
+
"151873": {
|
| 1846 |
+
"content": "|<EXTRA_TOKENS_204>|",
|
| 1847 |
+
"lstrip": false,
|
| 1848 |
+
"normalized": false,
|
| 1849 |
+
"rstrip": false,
|
| 1850 |
+
"single_word": false,
|
| 1851 |
+
"special": true
|
| 1852 |
+
},
|
| 1853 |
+
"151874": {
|
| 1854 |
+
"content": "|<EXTRA_TOKENS_205>|",
|
| 1855 |
+
"lstrip": false,
|
| 1856 |
+
"normalized": false,
|
| 1857 |
+
"rstrip": false,
|
| 1858 |
+
"single_word": false,
|
| 1859 |
+
"special": true
|
| 1860 |
+
},
|
| 1861 |
+
"151875": {
|
| 1862 |
+
"content": "|<EXTRA_TOKENS_206>|",
|
| 1863 |
+
"lstrip": false,
|
| 1864 |
+
"normalized": false,
|
| 1865 |
+
"rstrip": false,
|
| 1866 |
+
"single_word": false,
|
| 1867 |
+
"special": true
|
| 1868 |
+
},
|
| 1869 |
+
"151876": {
|
| 1870 |
+
"content": "|<EXTRA_TOKENS_207>|",
|
| 1871 |
+
"lstrip": false,
|
| 1872 |
+
"normalized": false,
|
| 1873 |
+
"rstrip": false,
|
| 1874 |
+
"single_word": false,
|
| 1875 |
+
"special": true
|
| 1876 |
+
},
|
| 1877 |
+
"151877": {
|
| 1878 |
+
"content": "|<EXTRA_TOKENS_208>|",
|
| 1879 |
+
"lstrip": false,
|
| 1880 |
+
"normalized": false,
|
| 1881 |
+
"rstrip": false,
|
| 1882 |
+
"single_word": false,
|
| 1883 |
+
"special": true
|
| 1884 |
+
},
|
| 1885 |
+
"151878": {
|
| 1886 |
+
"content": "|<EXTRA_TOKENS_209>|",
|
| 1887 |
+
"lstrip": false,
|
| 1888 |
+
"normalized": false,
|
| 1889 |
+
"rstrip": false,
|
| 1890 |
+
"single_word": false,
|
| 1891 |
+
"special": true
|
| 1892 |
+
},
|
| 1893 |
+
"151879": {
|
| 1894 |
+
"content": "|<EXTRA_TOKENS_210>|",
|
| 1895 |
+
"lstrip": false,
|
| 1896 |
+
"normalized": false,
|
| 1897 |
+
"rstrip": false,
|
| 1898 |
+
"single_word": false,
|
| 1899 |
+
"special": true
|
| 1900 |
+
},
|
| 1901 |
+
"151880": {
|
| 1902 |
+
"content": "|<EXTRA_TOKENS_211>|",
|
| 1903 |
+
"lstrip": false,
|
| 1904 |
+
"normalized": false,
|
| 1905 |
+
"rstrip": false,
|
| 1906 |
+
"single_word": false,
|
| 1907 |
+
"special": true
|
| 1908 |
+
},
|
| 1909 |
+
"151881": {
|
| 1910 |
+
"content": "|<EXTRA_TOKENS_212>|",
|
| 1911 |
+
"lstrip": false,
|
| 1912 |
+
"normalized": false,
|
| 1913 |
+
"rstrip": false,
|
| 1914 |
+
"single_word": false,
|
| 1915 |
+
"special": true
|
| 1916 |
+
},
|
| 1917 |
+
"151882": {
|
| 1918 |
+
"content": "|<EXTRA_TOKENS_213>|",
|
| 1919 |
+
"lstrip": false,
|
| 1920 |
+
"normalized": false,
|
| 1921 |
+
"rstrip": false,
|
| 1922 |
+
"single_word": false,
|
| 1923 |
+
"special": true
|
| 1924 |
+
},
|
| 1925 |
+
"151883": {
|
| 1926 |
+
"content": "|<EXTRA_TOKENS_214>|",
|
| 1927 |
+
"lstrip": false,
|
| 1928 |
+
"normalized": false,
|
| 1929 |
+
"rstrip": false,
|
| 1930 |
+
"single_word": false,
|
| 1931 |
+
"special": true
|
| 1932 |
+
},
|
| 1933 |
+
"151884": {
|
| 1934 |
+
"content": "|<EXTRA_TOKENS_215>|",
|
| 1935 |
+
"lstrip": false,
|
| 1936 |
+
"normalized": false,
|
| 1937 |
+
"rstrip": false,
|
| 1938 |
+
"single_word": false,
|
| 1939 |
+
"special": true
|
| 1940 |
+
},
|
| 1941 |
+
"151885": {
|
| 1942 |
+
"content": "|<EXTRA_TOKENS_216>|",
|
| 1943 |
+
"lstrip": false,
|
| 1944 |
+
"normalized": false,
|
| 1945 |
+
"rstrip": false,
|
| 1946 |
+
"single_word": false,
|
| 1947 |
+
"special": true
|
| 1948 |
+
},
|
| 1949 |
+
"151886": {
|
| 1950 |
+
"content": "|<EXTRA_TOKENS_217>|",
|
| 1951 |
+
"lstrip": false,
|
| 1952 |
+
"normalized": false,
|
| 1953 |
+
"rstrip": false,
|
| 1954 |
+
"single_word": false,
|
| 1955 |
+
"special": true
|
| 1956 |
+
},
|
| 1957 |
+
"151887": {
|
| 1958 |
+
"content": "|<EXTRA_TOKENS_218>|",
|
| 1959 |
+
"lstrip": false,
|
| 1960 |
+
"normalized": false,
|
| 1961 |
+
"rstrip": false,
|
| 1962 |
+
"single_word": false,
|
| 1963 |
+
"special": true
|
| 1964 |
+
},
|
| 1965 |
+
"151888": {
|
| 1966 |
+
"content": "|<EXTRA_TOKENS_219>|",
|
| 1967 |
+
"lstrip": false,
|
| 1968 |
+
"normalized": false,
|
| 1969 |
+
"rstrip": false,
|
| 1970 |
+
"single_word": false,
|
| 1971 |
+
"special": true
|
| 1972 |
+
},
|
| 1973 |
+
"151889": {
|
| 1974 |
+
"content": "|<EXTRA_TOKENS_220>|",
|
| 1975 |
+
"lstrip": false,
|
| 1976 |
+
"normalized": false,
|
| 1977 |
+
"rstrip": false,
|
| 1978 |
+
"single_word": false,
|
| 1979 |
+
"special": true
|
| 1980 |
+
},
|
| 1981 |
+
"151890": {
|
| 1982 |
+
"content": "|<EXTRA_TOKENS_221>|",
|
| 1983 |
+
"lstrip": false,
|
| 1984 |
+
"normalized": false,
|
| 1985 |
+
"rstrip": false,
|
| 1986 |
+
"single_word": false,
|
| 1987 |
+
"special": true
|
| 1988 |
+
},
|
| 1989 |
+
"151891": {
|
| 1990 |
+
"content": "|<EXTRA_TOKENS_222>|",
|
| 1991 |
+
"lstrip": false,
|
| 1992 |
+
"normalized": false,
|
| 1993 |
+
"rstrip": false,
|
| 1994 |
+
"single_word": false,
|
| 1995 |
+
"special": true
|
| 1996 |
+
},
|
| 1997 |
+
"151892": {
|
| 1998 |
+
"content": "|<EXTRA_TOKENS_223>|",
|
| 1999 |
+
"lstrip": false,
|
| 2000 |
+
"normalized": false,
|
| 2001 |
+
"rstrip": false,
|
| 2002 |
+
"single_word": false,
|
| 2003 |
+
"special": true
|
| 2004 |
+
},
|
| 2005 |
+
"151893": {
|
| 2006 |
+
"content": "|<EXTRA_TOKENS_224>|",
|
| 2007 |
+
"lstrip": false,
|
| 2008 |
+
"normalized": false,
|
| 2009 |
+
"rstrip": false,
|
| 2010 |
+
"single_word": false,
|
| 2011 |
+
"special": true
|
| 2012 |
+
},
|
| 2013 |
+
"151894": {
|
| 2014 |
+
"content": "|<EXTRA_TOKENS_225>|",
|
| 2015 |
+
"lstrip": false,
|
| 2016 |
+
"normalized": false,
|
| 2017 |
+
"rstrip": false,
|
| 2018 |
+
"single_word": false,
|
| 2019 |
+
"special": true
|
| 2020 |
+
},
|
| 2021 |
+
"151895": {
|
| 2022 |
+
"content": "|<EXTRA_TOKENS_226>|",
|
| 2023 |
+
"lstrip": false,
|
| 2024 |
+
"normalized": false,
|
| 2025 |
+
"rstrip": false,
|
| 2026 |
+
"single_word": false,
|
| 2027 |
+
"special": true
|
| 2028 |
+
},
|
| 2029 |
+
"151896": {
|
| 2030 |
+
"content": "|<EXTRA_TOKENS_227>|",
|
| 2031 |
+
"lstrip": false,
|
| 2032 |
+
"normalized": false,
|
| 2033 |
+
"rstrip": false,
|
| 2034 |
+
"single_word": false,
|
| 2035 |
+
"special": true
|
| 2036 |
+
},
|
| 2037 |
+
"151897": {
|
| 2038 |
+
"content": "|<EXTRA_TOKENS_228>|",
|
| 2039 |
+
"lstrip": false,
|
| 2040 |
+
"normalized": false,
|
| 2041 |
+
"rstrip": false,
|
| 2042 |
+
"single_word": false,
|
| 2043 |
+
"special": true
|
| 2044 |
+
},
|
| 2045 |
+
"151898": {
|
| 2046 |
+
"content": "|<EXTRA_TOKENS_229>|",
|
| 2047 |
+
"lstrip": false,
|
| 2048 |
+
"normalized": false,
|
| 2049 |
+
"rstrip": false,
|
| 2050 |
+
"single_word": false,
|
| 2051 |
+
"special": true
|
| 2052 |
+
},
|
| 2053 |
+
"151899": {
|
| 2054 |
+
"content": "|<EXTRA_TOKENS_230>|",
|
| 2055 |
+
"lstrip": false,
|
| 2056 |
+
"normalized": false,
|
| 2057 |
+
"rstrip": false,
|
| 2058 |
+
"single_word": false,
|
| 2059 |
+
"special": true
|
| 2060 |
+
},
|
| 2061 |
+
"151900": {
|
| 2062 |
+
"content": "|<EXTRA_TOKENS_231>|",
|
| 2063 |
+
"lstrip": false,
|
| 2064 |
+
"normalized": false,
|
| 2065 |
+
"rstrip": false,
|
| 2066 |
+
"single_word": false,
|
| 2067 |
+
"special": true
|
| 2068 |
+
},
|
| 2069 |
+
"151901": {
|
| 2070 |
+
"content": "|<EXTRA_TOKENS_232>|",
|
| 2071 |
+
"lstrip": false,
|
| 2072 |
+
"normalized": false,
|
| 2073 |
+
"rstrip": false,
|
| 2074 |
+
"single_word": false,
|
| 2075 |
+
"special": true
|
| 2076 |
+
},
|
| 2077 |
+
"151902": {
|
| 2078 |
+
"content": "|<EXTRA_TOKENS_233>|",
|
| 2079 |
+
"lstrip": false,
|
| 2080 |
+
"normalized": false,
|
| 2081 |
+
"rstrip": false,
|
| 2082 |
+
"single_word": false,
|
| 2083 |
+
"special": true
|
| 2084 |
+
},
|
| 2085 |
+
"151903": {
|
| 2086 |
+
"content": "|<EXTRA_TOKENS_234>|",
|
| 2087 |
+
"lstrip": false,
|
| 2088 |
+
"normalized": false,
|
| 2089 |
+
"rstrip": false,
|
| 2090 |
+
"single_word": false,
|
| 2091 |
+
"special": true
|
| 2092 |
+
},
|
| 2093 |
+
"151904": {
|
| 2094 |
+
"content": "|<EXTRA_TOKENS_235>|",
|
| 2095 |
+
"lstrip": false,
|
| 2096 |
+
"normalized": false,
|
| 2097 |
+
"rstrip": false,
|
| 2098 |
+
"single_word": false,
|
| 2099 |
+
"special": true
|
| 2100 |
+
},
|
| 2101 |
+
"151905": {
|
| 2102 |
+
"content": "|<EXTRA_TOKENS_236>|",
|
| 2103 |
+
"lstrip": false,
|
| 2104 |
+
"normalized": false,
|
| 2105 |
+
"rstrip": false,
|
| 2106 |
+
"single_word": false,
|
| 2107 |
+
"special": true
|
| 2108 |
+
},
|
| 2109 |
+
"151906": {
|
| 2110 |
+
"content": "|<EXTRA_TOKENS_237>|",
|
| 2111 |
+
"lstrip": false,
|
| 2112 |
+
"normalized": false,
|
| 2113 |
+
"rstrip": false,
|
| 2114 |
+
"single_word": false,
|
| 2115 |
+
"special": true
|
| 2116 |
+
},
|
| 2117 |
+
"151907": {
|
| 2118 |
+
"content": "|<EXTRA_TOKENS_238>|",
|
| 2119 |
+
"lstrip": false,
|
| 2120 |
+
"normalized": false,
|
| 2121 |
+
"rstrip": false,
|
| 2122 |
+
"single_word": false,
|
| 2123 |
+
"special": true
|
| 2124 |
+
},
|
| 2125 |
+
"151908": {
|
| 2126 |
+
"content": "|<EXTRA_TOKENS_239>|",
|
| 2127 |
+
"lstrip": false,
|
| 2128 |
+
"normalized": false,
|
| 2129 |
+
"rstrip": false,
|
| 2130 |
+
"single_word": false,
|
| 2131 |
+
"special": true
|
| 2132 |
+
},
|
| 2133 |
+
"151909": {
|
| 2134 |
+
"content": "|<EXTRA_TOKENS_240>|",
|
| 2135 |
+
"lstrip": false,
|
| 2136 |
+
"normalized": false,
|
| 2137 |
+
"rstrip": false,
|
| 2138 |
+
"single_word": false,
|
| 2139 |
+
"special": true
|
| 2140 |
+
},
|
| 2141 |
+
"151910": {
|
| 2142 |
+
"content": "|<EXTRA_TOKENS_241>|",
|
| 2143 |
+
"lstrip": false,
|
| 2144 |
+
"normalized": false,
|
| 2145 |
+
"rstrip": false,
|
| 2146 |
+
"single_word": false,
|
| 2147 |
+
"special": true
|
| 2148 |
+
},
|
| 2149 |
+
"151911": {
|
| 2150 |
+
"content": "|<EXTRA_TOKENS_242>|",
|
| 2151 |
+
"lstrip": false,
|
| 2152 |
+
"normalized": false,
|
| 2153 |
+
"rstrip": false,
|
| 2154 |
+
"single_word": false,
|
| 2155 |
+
"special": true
|
| 2156 |
+
},
|
| 2157 |
+
"151912": {
|
| 2158 |
+
"content": "|<EXTRA_TOKENS_243>|",
|
| 2159 |
+
"lstrip": false,
|
| 2160 |
+
"normalized": false,
|
| 2161 |
+
"rstrip": false,
|
| 2162 |
+
"single_word": false,
|
| 2163 |
+
"special": true
|
| 2164 |
+
},
|
| 2165 |
+
"151913": {
|
| 2166 |
+
"content": "|<EXTRA_TOKENS_244>|",
|
| 2167 |
+
"lstrip": false,
|
| 2168 |
+
"normalized": false,
|
| 2169 |
+
"rstrip": false,
|
| 2170 |
+
"single_word": false,
|
| 2171 |
+
"special": true
|
| 2172 |
+
},
|
| 2173 |
+
"151914": {
|
| 2174 |
+
"content": "|<EXTRA_TOKENS_245>|",
|
| 2175 |
+
"lstrip": false,
|
| 2176 |
+
"normalized": false,
|
| 2177 |
+
"rstrip": false,
|
| 2178 |
+
"single_word": false,
|
| 2179 |
+
"special": true
|
| 2180 |
+
},
|
| 2181 |
+
"151915": {
|
| 2182 |
+
"content": "|<EXTRA_TOKENS_246>|",
|
| 2183 |
+
"lstrip": false,
|
| 2184 |
+
"normalized": false,
|
| 2185 |
+
"rstrip": false,
|
| 2186 |
+
"single_word": false,
|
| 2187 |
+
"special": true
|
| 2188 |
+
},
|
| 2189 |
+
"151916": {
|
| 2190 |
+
"content": "|<EXTRA_TOKENS_247>|",
|
| 2191 |
+
"lstrip": false,
|
| 2192 |
+
"normalized": false,
|
| 2193 |
+
"rstrip": false,
|
| 2194 |
+
"single_word": false,
|
| 2195 |
+
"special": true
|
| 2196 |
+
},
|
| 2197 |
+
"151917": {
|
| 2198 |
+
"content": "|<EXTRA_TOKENS_248>|",
|
| 2199 |
+
"lstrip": false,
|
| 2200 |
+
"normalized": false,
|
| 2201 |
+
"rstrip": false,
|
| 2202 |
+
"single_word": false,
|
| 2203 |
+
"special": true
|
| 2204 |
+
},
|
| 2205 |
+
"151918": {
|
| 2206 |
+
"content": "|<EXTRA_TOKENS_249>|",
|
| 2207 |
+
"lstrip": false,
|
| 2208 |
+
"normalized": false,
|
| 2209 |
+
"rstrip": false,
|
| 2210 |
+
"single_word": false,
|
| 2211 |
+
"special": true
|
| 2212 |
+
},
|
| 2213 |
+
"151919": {
|
| 2214 |
+
"content": "|<EXTRA_TOKENS_250>|",
|
| 2215 |
+
"lstrip": false,
|
| 2216 |
+
"normalized": false,
|
| 2217 |
+
"rstrip": false,
|
| 2218 |
+
"single_word": false,
|
| 2219 |
+
"special": true
|
| 2220 |
+
},
|
| 2221 |
+
"151920": {
|
| 2222 |
+
"content": "|<EXTRA_TOKENS_251>|",
|
| 2223 |
+
"lstrip": false,
|
| 2224 |
+
"normalized": false,
|
| 2225 |
+
"rstrip": false,
|
| 2226 |
+
"single_word": false,
|
| 2227 |
+
"special": true
|
| 2228 |
+
},
|
| 2229 |
+
"151921": {
|
| 2230 |
+
"content": "|<EXTRA_TOKENS_252>|",
|
| 2231 |
+
"lstrip": false,
|
| 2232 |
+
"normalized": false,
|
| 2233 |
+
"rstrip": false,
|
| 2234 |
+
"single_word": false,
|
| 2235 |
+
"special": true
|
| 2236 |
+
},
|
| 2237 |
+
"151922": {
|
| 2238 |
+
"content": "|<EXTRA_TOKENS_253>|",
|
| 2239 |
+
"lstrip": false,
|
| 2240 |
+
"normalized": false,
|
| 2241 |
+
"rstrip": false,
|
| 2242 |
+
"single_word": false,
|
| 2243 |
+
"special": true
|
| 2244 |
+
},
|
| 2245 |
+
"151923": {
|
| 2246 |
+
"content": "|<EXTRA_TOKENS_254>|",
|
| 2247 |
+
"lstrip": false,
|
| 2248 |
+
"normalized": false,
|
| 2249 |
+
"rstrip": false,
|
| 2250 |
+
"single_word": false,
|
| 2251 |
+
"special": true
|
| 2252 |
+
},
|
| 2253 |
+
"151924": {
|
| 2254 |
+
"content": "|<EXTRA_TOKENS_255>|",
|
| 2255 |
+
"lstrip": false,
|
| 2256 |
+
"normalized": false,
|
| 2257 |
+
"rstrip": false,
|
| 2258 |
+
"single_word": false,
|
| 2259 |
+
"special": true
|
| 2260 |
+
},
|
| 2261 |
+
"151925": {
|
| 2262 |
+
"content": "|<EXTRA_TOKENS_256>|",
|
| 2263 |
+
"lstrip": false,
|
| 2264 |
+
"normalized": false,
|
| 2265 |
+
"rstrip": false,
|
| 2266 |
+
"single_word": false,
|
| 2267 |
+
"special": true
|
| 2268 |
+
},
|
| 2269 |
+
"151926": {
|
| 2270 |
+
"content": "|<EXTRA_TOKENS_257>|",
|
| 2271 |
+
"lstrip": false,
|
| 2272 |
+
"normalized": false,
|
| 2273 |
+
"rstrip": false,
|
| 2274 |
+
"single_word": false,
|
| 2275 |
+
"special": true
|
| 2276 |
+
},
|
| 2277 |
+
"151927": {
|
| 2278 |
+
"content": "|<EXTRA_TOKENS_258>|",
|
| 2279 |
+
"lstrip": false,
|
| 2280 |
+
"normalized": false,
|
| 2281 |
+
"rstrip": false,
|
| 2282 |
+
"single_word": false,
|
| 2283 |
+
"special": true
|
| 2284 |
+
},
|
| 2285 |
+
"151928": {
|
| 2286 |
+
"content": "|<EXTRA_TOKENS_259>|",
|
| 2287 |
+
"lstrip": false,
|
| 2288 |
+
"normalized": false,
|
| 2289 |
+
"rstrip": false,
|
| 2290 |
+
"single_word": false,
|
| 2291 |
+
"special": true
|
| 2292 |
+
},
|
| 2293 |
+
"151929": {
|
| 2294 |
+
"content": "|<EXTRA_TOKENS_260>|",
|
| 2295 |
+
"lstrip": false,
|
| 2296 |
+
"normalized": false,
|
| 2297 |
+
"rstrip": false,
|
| 2298 |
+
"single_word": false,
|
| 2299 |
+
"special": true
|
| 2300 |
+
},
|
| 2301 |
+
"151930": {
|
| 2302 |
+
"content": "|<EXTRA_TOKENS_261>|",
|
| 2303 |
+
"lstrip": false,
|
| 2304 |
+
"normalized": false,
|
| 2305 |
+
"rstrip": false,
|
| 2306 |
+
"single_word": false,
|
| 2307 |
+
"special": true
|
| 2308 |
+
},
|
| 2309 |
+
"151931": {
|
| 2310 |
+
"content": "|<EXTRA_TOKENS_262>|",
|
| 2311 |
+
"lstrip": false,
|
| 2312 |
+
"normalized": false,
|
| 2313 |
+
"rstrip": false,
|
| 2314 |
+
"single_word": false,
|
| 2315 |
+
"special": true
|
| 2316 |
+
},
|
| 2317 |
+
"151932": {
|
| 2318 |
+
"content": "|<EXTRA_TOKENS_263>|",
|
| 2319 |
+
"lstrip": false,
|
| 2320 |
+
"normalized": false,
|
| 2321 |
+
"rstrip": false,
|
| 2322 |
+
"single_word": false,
|
| 2323 |
+
"special": true
|
| 2324 |
+
},
|
| 2325 |
+
"151933": {
|
| 2326 |
+
"content": "|<EXTRA_TOKENS_264>|",
|
| 2327 |
+
"lstrip": false,
|
| 2328 |
+
"normalized": false,
|
| 2329 |
+
"rstrip": false,
|
| 2330 |
+
"single_word": false,
|
| 2331 |
+
"special": true
|
| 2332 |
+
},
|
| 2333 |
+
"151934": {
|
| 2334 |
+
"content": "|<EXTRA_TOKENS_265>|",
|
| 2335 |
+
"lstrip": false,
|
| 2336 |
+
"normalized": false,
|
| 2337 |
+
"rstrip": false,
|
| 2338 |
+
"single_word": false,
|
| 2339 |
+
"special": true
|
| 2340 |
+
},
|
| 2341 |
+
"151935": {
|
| 2342 |
+
"content": "|<EXTRA_TOKENS_266>|",
|
| 2343 |
+
"lstrip": false,
|
| 2344 |
+
"normalized": false,
|
| 2345 |
+
"rstrip": false,
|
| 2346 |
+
"single_word": false,
|
| 2347 |
+
"special": true
|
| 2348 |
+
},
|
| 2349 |
+
"151936": {
|
| 2350 |
+
"content": "<im_start>",
|
| 2351 |
+
"lstrip": false,
|
| 2352 |
+
"normalized": false,
|
| 2353 |
+
"rstrip": false,
|
| 2354 |
+
"single_word": false,
|
| 2355 |
+
"special": true
|
| 2356 |
+
},
|
| 2357 |
+
"151937": {
|
| 2358 |
+
"content": "<im_end>",
|
| 2359 |
+
"lstrip": false,
|
| 2360 |
+
"normalized": false,
|
| 2361 |
+
"rstrip": false,
|
| 2362 |
+
"single_word": false,
|
| 2363 |
+
"special": true
|
| 2364 |
+
},
|
| 2365 |
+
"151938": {
|
| 2366 |
+
"content": "<im_patch>",
|
| 2367 |
+
"lstrip": false,
|
| 2368 |
+
"normalized": false,
|
| 2369 |
+
"rstrip": false,
|
| 2370 |
+
"single_word": false,
|
| 2371 |
+
"special": true
|
| 2372 |
+
},
|
| 2373 |
+
"151939": {
|
| 2374 |
+
"content": "<im_col>",
|
| 2375 |
+
"lstrip": false,
|
| 2376 |
+
"normalized": false,
|
| 2377 |
+
"rstrip": false,
|
| 2378 |
+
"single_word": false,
|
| 2379 |
+
"special": true
|
| 2380 |
+
},
|
| 2381 |
+
"151940": {
|
| 2382 |
+
"content": "<low_res_im_start>",
|
| 2383 |
+
"lstrip": false,
|
| 2384 |
+
"normalized": false,
|
| 2385 |
+
"rstrip": false,
|
| 2386 |
+
"single_word": false,
|
| 2387 |
+
"special": true
|
| 2388 |
+
},
|
| 2389 |
+
"151941": {
|
| 2390 |
+
"content": "<|image|>",
|
| 2391 |
+
"lstrip": false,
|
| 2392 |
+
"normalized": false,
|
| 2393 |
+
"rstrip": false,
|
| 2394 |
+
"single_word": false,
|
| 2395 |
+
"special": true
|
| 2396 |
+
},
|
| 2397 |
+
"151942": {
|
| 2398 |
+
"content": "<im_low>",
|
| 2399 |
+
"lstrip": false,
|
| 2400 |
+
"normalized": false,
|
| 2401 |
+
"rstrip": false,
|
| 2402 |
+
"single_word": false,
|
| 2403 |
+
"special": true
|
| 2404 |
+
},
|
| 2405 |
+
"151943": {
|
| 2406 |
+
"content": "<frame_start>",
|
| 2407 |
+
"lstrip": false,
|
| 2408 |
+
"normalized": false,
|
| 2409 |
+
"rstrip": false,
|
| 2410 |
+
"single_word": false,
|
| 2411 |
+
"special": true
|
| 2412 |
+
},
|
| 2413 |
+
"151944": {
|
| 2414 |
+
"content": "<frame_end>",
|
| 2415 |
+
"lstrip": false,
|
| 2416 |
+
"normalized": false,
|
| 2417 |
+
"rstrip": false,
|
| 2418 |
+
"single_word": false,
|
| 2419 |
+
"special": true
|
| 2420 |
+
},
|
| 2421 |
+
"151945": {
|
| 2422 |
+
"content": "<|video|>",
|
| 2423 |
+
"lstrip": false,
|
| 2424 |
+
"normalized": false,
|
| 2425 |
+
"rstrip": false,
|
| 2426 |
+
"single_word": false,
|
| 2427 |
+
"special": true
|
| 2428 |
+
}
|
| 2429 |
+
},
|
| 2430 |
+
"additional_special_tokens": [
|
| 2431 |
+
"|<EXTRA_TOKENS_0>|",
|
| 2432 |
+
"|<EXTRA_TOKENS_1>|",
|
| 2433 |
+
"|<EXTRA_TOKENS_2>|",
|
| 2434 |
+
"|<EXTRA_TOKENS_3>|",
|
| 2435 |
+
"|<EXTRA_TOKENS_4>|",
|
| 2436 |
+
"|<EXTRA_TOKENS_5>|",
|
| 2437 |
+
"|<EXTRA_TOKENS_6>|",
|
| 2438 |
+
"|<EXTRA_TOKENS_7>|",
|
| 2439 |
+
"|<EXTRA_TOKENS_8>|",
|
| 2440 |
+
"|<EXTRA_TOKENS_9>|",
|
| 2441 |
+
"|<EXTRA_TOKENS_10>|",
|
| 2442 |
+
"|<EXTRA_TOKENS_11>|",
|
| 2443 |
+
"|<EXTRA_TOKENS_12>|",
|
| 2444 |
+
"|<EXTRA_TOKENS_13>|",
|
| 2445 |
+
"|<EXTRA_TOKENS_14>|",
|
| 2446 |
+
"|<EXTRA_TOKENS_15>|",
|
| 2447 |
+
"|<EXTRA_TOKENS_16>|",
|
| 2448 |
+
"|<EXTRA_TOKENS_17>|",
|
| 2449 |
+
"|<EXTRA_TOKENS_18>|",
|
| 2450 |
+
"|<EXTRA_TOKENS_19>|",
|
| 2451 |
+
"|<EXTRA_TOKENS_20>|",
|
| 2452 |
+
"|<EXTRA_TOKENS_21>|",
|
| 2453 |
+
"|<EXTRA_TOKENS_22>|",
|
| 2454 |
+
"|<EXTRA_TOKENS_23>|",
|
| 2455 |
+
"|<EXTRA_TOKENS_24>|",
|
| 2456 |
+
"|<EXTRA_TOKENS_25>|",
|
| 2457 |
+
"|<EXTRA_TOKENS_26>|",
|
| 2458 |
+
"|<EXTRA_TOKENS_27>|",
|
| 2459 |
+
"|<EXTRA_TOKENS_28>|",
|
| 2460 |
+
"|<EXTRA_TOKENS_29>|",
|
| 2461 |
+
"|<EXTRA_TOKENS_30>|",
|
| 2462 |
+
"|<EXTRA_TOKENS_31>|",
|
| 2463 |
+
"|<EXTRA_TOKENS_32>|",
|
| 2464 |
+
"|<EXTRA_TOKENS_33>|",
|
| 2465 |
+
"|<EXTRA_TOKENS_34>|",
|
| 2466 |
+
"|<EXTRA_TOKENS_35>|",
|
| 2467 |
+
"|<EXTRA_TOKENS_36>|",
|
| 2468 |
+
"|<EXTRA_TOKENS_37>|",
|
| 2469 |
+
"|<EXTRA_TOKENS_38>|",
|
| 2470 |
+
"|<EXTRA_TOKENS_39>|",
|
| 2471 |
+
"|<EXTRA_TOKENS_40>|",
|
| 2472 |
+
"|<EXTRA_TOKENS_41>|",
|
| 2473 |
+
"|<EXTRA_TOKENS_42>|",
|
| 2474 |
+
"|<EXTRA_TOKENS_43>|",
|
| 2475 |
+
"|<EXTRA_TOKENS_44>|",
|
| 2476 |
+
"|<EXTRA_TOKENS_45>|",
|
| 2477 |
+
"|<EXTRA_TOKENS_46>|",
|
| 2478 |
+
"|<EXTRA_TOKENS_47>|",
|
| 2479 |
+
"|<EXTRA_TOKENS_48>|",
|
| 2480 |
+
"|<EXTRA_TOKENS_49>|",
|
| 2481 |
+
"|<EXTRA_TOKENS_50>|",
|
| 2482 |
+
"|<EXTRA_TOKENS_51>|",
|
| 2483 |
+
"|<EXTRA_TOKENS_52>|",
|
| 2484 |
+
"|<EXTRA_TOKENS_53>|",
|
| 2485 |
+
"|<EXTRA_TOKENS_54>|",
|
| 2486 |
+
"|<EXTRA_TOKENS_55>|",
|
| 2487 |
+
"|<EXTRA_TOKENS_56>|",
|
| 2488 |
+
"|<EXTRA_TOKENS_57>|",
|
| 2489 |
+
"|<EXTRA_TOKENS_58>|",
|
| 2490 |
+
"|<EXTRA_TOKENS_59>|",
|
| 2491 |
+
"|<EXTRA_TOKENS_60>|",
|
| 2492 |
+
"|<EXTRA_TOKENS_61>|",
|
| 2493 |
+
"|<EXTRA_TOKENS_62>|",
|
| 2494 |
+
"|<EXTRA_TOKENS_63>|",
|
| 2495 |
+
"|<EXTRA_TOKENS_64>|",
|
| 2496 |
+
"|<EXTRA_TOKENS_65>|",
|
| 2497 |
+
"|<EXTRA_TOKENS_66>|",
|
| 2498 |
+
"|<EXTRA_TOKENS_67>|",
|
| 2499 |
+
"|<EXTRA_TOKENS_68>|",
|
| 2500 |
+
"|<EXTRA_TOKENS_69>|",
|
| 2501 |
+
"|<EXTRA_TOKENS_70>|",
|
| 2502 |
+
"|<EXTRA_TOKENS_71>|",
|
| 2503 |
+
"|<EXTRA_TOKENS_72>|",
|
| 2504 |
+
"|<EXTRA_TOKENS_73>|",
|
| 2505 |
+
"|<EXTRA_TOKENS_74>|",
|
| 2506 |
+
"|<EXTRA_TOKENS_75>|",
|
| 2507 |
+
"|<EXTRA_TOKENS_76>|",
|
| 2508 |
+
"|<EXTRA_TOKENS_77>|",
|
| 2509 |
+
"|<EXTRA_TOKENS_78>|",
|
| 2510 |
+
"|<EXTRA_TOKENS_79>|",
|
| 2511 |
+
"|<EXTRA_TOKENS_80>|",
|
| 2512 |
+
"|<EXTRA_TOKENS_81>|",
|
| 2513 |
+
"|<EXTRA_TOKENS_82>|",
|
| 2514 |
+
"|<EXTRA_TOKENS_83>|",
|
| 2515 |
+
"|<EXTRA_TOKENS_84>|",
|
| 2516 |
+
"|<EXTRA_TOKENS_85>|",
|
| 2517 |
+
"|<EXTRA_TOKENS_86>|",
|
| 2518 |
+
"|<EXTRA_TOKENS_87>|",
|
| 2519 |
+
"|<EXTRA_TOKENS_88>|",
|
| 2520 |
+
"|<EXTRA_TOKENS_89>|",
|
| 2521 |
+
"|<EXTRA_TOKENS_90>|",
|
| 2522 |
+
"|<EXTRA_TOKENS_91>|",
|
| 2523 |
+
"|<EXTRA_TOKENS_92>|",
|
| 2524 |
+
"|<EXTRA_TOKENS_93>|",
|
| 2525 |
+
"|<EXTRA_TOKENS_94>|",
|
| 2526 |
+
"|<EXTRA_TOKENS_95>|",
|
| 2527 |
+
"|<EXTRA_TOKENS_96>|",
|
| 2528 |
+
"|<EXTRA_TOKENS_97>|",
|
| 2529 |
+
"|<EXTRA_TOKENS_98>|",
|
| 2530 |
+
"|<EXTRA_TOKENS_99>|",
|
| 2531 |
+
"|<EXTRA_TOKENS_100>|",
|
| 2532 |
+
"|<EXTRA_TOKENS_101>|",
|
| 2533 |
+
"|<EXTRA_TOKENS_102>|",
|
| 2534 |
+
"|<EXTRA_TOKENS_103>|",
|
| 2535 |
+
"|<EXTRA_TOKENS_104>|",
|
| 2536 |
+
"|<EXTRA_TOKENS_105>|",
|
| 2537 |
+
"|<EXTRA_TOKENS_106>|",
|
| 2538 |
+
"|<EXTRA_TOKENS_107>|",
|
| 2539 |
+
"|<EXTRA_TOKENS_108>|",
|
| 2540 |
+
"|<EXTRA_TOKENS_109>|",
|
| 2541 |
+
"|<EXTRA_TOKENS_110>|",
|
| 2542 |
+
"|<EXTRA_TOKENS_111>|",
|
| 2543 |
+
"|<EXTRA_TOKENS_112>|",
|
| 2544 |
+
"|<EXTRA_TOKENS_113>|",
|
| 2545 |
+
"|<EXTRA_TOKENS_114>|",
|
| 2546 |
+
"|<EXTRA_TOKENS_115>|",
|
| 2547 |
+
"|<EXTRA_TOKENS_116>|",
|
| 2548 |
+
"|<EXTRA_TOKENS_117>|",
|
| 2549 |
+
"|<EXTRA_TOKENS_118>|",
|
| 2550 |
+
"|<EXTRA_TOKENS_119>|",
|
| 2551 |
+
"|<EXTRA_TOKENS_120>|",
|
| 2552 |
+
"|<EXTRA_TOKENS_121>|",
|
| 2553 |
+
"|<EXTRA_TOKENS_122>|",
|
| 2554 |
+
"|<EXTRA_TOKENS_123>|",
|
| 2555 |
+
"|<EXTRA_TOKENS_124>|",
|
| 2556 |
+
"|<EXTRA_TOKENS_125>|",
|
| 2557 |
+
"|<EXTRA_TOKENS_126>|",
|
| 2558 |
+
"|<EXTRA_TOKENS_127>|",
|
| 2559 |
+
"|<EXTRA_TOKENS_128>|",
|
| 2560 |
+
"|<EXTRA_TOKENS_129>|",
|
| 2561 |
+
"|<EXTRA_TOKENS_130>|",
|
| 2562 |
+
"|<EXTRA_TOKENS_131>|",
|
| 2563 |
+
"|<EXTRA_TOKENS_132>|",
|
| 2564 |
+
"|<EXTRA_TOKENS_133>|",
|
| 2565 |
+
"|<EXTRA_TOKENS_134>|",
|
| 2566 |
+
"|<EXTRA_TOKENS_135>|",
|
| 2567 |
+
"|<EXTRA_TOKENS_136>|",
|
| 2568 |
+
"|<EXTRA_TOKENS_137>|",
|
| 2569 |
+
"|<EXTRA_TOKENS_138>|",
|
| 2570 |
+
"|<EXTRA_TOKENS_139>|",
|
| 2571 |
+
"|<EXTRA_TOKENS_140>|",
|
| 2572 |
+
"|<EXTRA_TOKENS_141>|",
|
| 2573 |
+
"|<EXTRA_TOKENS_142>|",
|
| 2574 |
+
"|<EXTRA_TOKENS_143>|",
|
| 2575 |
+
"|<EXTRA_TOKENS_144>|",
|
| 2576 |
+
"|<EXTRA_TOKENS_145>|",
|
| 2577 |
+
"|<EXTRA_TOKENS_146>|",
|
| 2578 |
+
"|<EXTRA_TOKENS_147>|",
|
| 2579 |
+
"|<EXTRA_TOKENS_148>|",
|
| 2580 |
+
"|<EXTRA_TOKENS_149>|",
|
| 2581 |
+
"|<EXTRA_TOKENS_150>|",
|
| 2582 |
+
"|<EXTRA_TOKENS_151>|",
|
| 2583 |
+
"|<EXTRA_TOKENS_152>|",
|
| 2584 |
+
"|<EXTRA_TOKENS_153>|",
|
| 2585 |
+
"|<EXTRA_TOKENS_154>|",
|
| 2586 |
+
"|<EXTRA_TOKENS_155>|",
|
| 2587 |
+
"|<EXTRA_TOKENS_156>|",
|
| 2588 |
+
"|<EXTRA_TOKENS_157>|",
|
| 2589 |
+
"|<EXTRA_TOKENS_158>|",
|
| 2590 |
+
"|<EXTRA_TOKENS_159>|",
|
| 2591 |
+
"|<EXTRA_TOKENS_160>|",
|
| 2592 |
+
"|<EXTRA_TOKENS_161>|",
|
| 2593 |
+
"|<EXTRA_TOKENS_162>|",
|
| 2594 |
+
"|<EXTRA_TOKENS_163>|",
|
| 2595 |
+
"|<EXTRA_TOKENS_164>|",
|
| 2596 |
+
"|<EXTRA_TOKENS_165>|",
|
| 2597 |
+
"|<EXTRA_TOKENS_166>|",
|
| 2598 |
+
"|<EXTRA_TOKENS_167>|",
|
| 2599 |
+
"|<EXTRA_TOKENS_168>|",
|
| 2600 |
+
"|<EXTRA_TOKENS_169>|",
|
| 2601 |
+
"|<EXTRA_TOKENS_170>|",
|
| 2602 |
+
"|<EXTRA_TOKENS_171>|",
|
| 2603 |
+
"|<EXTRA_TOKENS_172>|",
|
| 2604 |
+
"|<EXTRA_TOKENS_173>|",
|
| 2605 |
+
"|<EXTRA_TOKENS_174>|",
|
| 2606 |
+
"|<EXTRA_TOKENS_175>|",
|
| 2607 |
+
"|<EXTRA_TOKENS_176>|",
|
| 2608 |
+
"|<EXTRA_TOKENS_177>|",
|
| 2609 |
+
"|<EXTRA_TOKENS_178>|",
|
| 2610 |
+
"|<EXTRA_TOKENS_179>|",
|
| 2611 |
+
"|<EXTRA_TOKENS_180>|",
|
| 2612 |
+
"|<EXTRA_TOKENS_181>|",
|
| 2613 |
+
"|<EXTRA_TOKENS_182>|",
|
| 2614 |
+
"|<EXTRA_TOKENS_183>|",
|
| 2615 |
+
"|<EXTRA_TOKENS_184>|",
|
| 2616 |
+
"|<EXTRA_TOKENS_185>|",
|
| 2617 |
+
"|<EXTRA_TOKENS_186>|",
|
| 2618 |
+
"|<EXTRA_TOKENS_187>|",
|
| 2619 |
+
"|<EXTRA_TOKENS_188>|",
|
| 2620 |
+
"|<EXTRA_TOKENS_189>|",
|
| 2621 |
+
"|<EXTRA_TOKENS_190>|",
|
| 2622 |
+
"|<EXTRA_TOKENS_191>|",
|
| 2623 |
+
"|<EXTRA_TOKENS_192>|",
|
| 2624 |
+
"|<EXTRA_TOKENS_193>|",
|
| 2625 |
+
"|<EXTRA_TOKENS_194>|",
|
| 2626 |
+
"|<EXTRA_TOKENS_195>|",
|
| 2627 |
+
"|<EXTRA_TOKENS_196>|",
|
| 2628 |
+
"|<EXTRA_TOKENS_197>|",
|
| 2629 |
+
"|<EXTRA_TOKENS_198>|",
|
| 2630 |
+
"|<EXTRA_TOKENS_199>|",
|
| 2631 |
+
"|<EXTRA_TOKENS_200>|",
|
| 2632 |
+
"|<EXTRA_TOKENS_201>|",
|
| 2633 |
+
"|<EXTRA_TOKENS_202>|",
|
| 2634 |
+
"|<EXTRA_TOKENS_203>|",
|
| 2635 |
+
"|<EXTRA_TOKENS_204>|",
|
| 2636 |
+
"|<EXTRA_TOKENS_205>|",
|
| 2637 |
+
"|<EXTRA_TOKENS_206>|",
|
| 2638 |
+
"|<EXTRA_TOKENS_207>|",
|
| 2639 |
+
"|<EXTRA_TOKENS_208>|",
|
| 2640 |
+
"|<EXTRA_TOKENS_209>|",
|
| 2641 |
+
"|<EXTRA_TOKENS_210>|",
|
| 2642 |
+
"|<EXTRA_TOKENS_211>|",
|
| 2643 |
+
"|<EXTRA_TOKENS_212>|",
|
| 2644 |
+
"|<EXTRA_TOKENS_213>|",
|
| 2645 |
+
"|<EXTRA_TOKENS_214>|",
|
| 2646 |
+
"|<EXTRA_TOKENS_215>|",
|
| 2647 |
+
"|<EXTRA_TOKENS_216>|",
|
| 2648 |
+
"|<EXTRA_TOKENS_217>|",
|
| 2649 |
+
"|<EXTRA_TOKENS_218>|",
|
| 2650 |
+
"|<EXTRA_TOKENS_219>|",
|
| 2651 |
+
"|<EXTRA_TOKENS_220>|",
|
| 2652 |
+
"|<EXTRA_TOKENS_221>|",
|
| 2653 |
+
"|<EXTRA_TOKENS_222>|",
|
| 2654 |
+
"|<EXTRA_TOKENS_223>|",
|
| 2655 |
+
"|<EXTRA_TOKENS_224>|",
|
| 2656 |
+
"|<EXTRA_TOKENS_225>|",
|
| 2657 |
+
"|<EXTRA_TOKENS_226>|",
|
| 2658 |
+
"|<EXTRA_TOKENS_227>|",
|
| 2659 |
+
"|<EXTRA_TOKENS_228>|",
|
| 2660 |
+
"|<EXTRA_TOKENS_229>|",
|
| 2661 |
+
"|<EXTRA_TOKENS_230>|",
|
| 2662 |
+
"|<EXTRA_TOKENS_231>|",
|
| 2663 |
+
"|<EXTRA_TOKENS_232>|",
|
| 2664 |
+
"|<EXTRA_TOKENS_233>|",
|
| 2665 |
+
"|<EXTRA_TOKENS_234>|",
|
| 2666 |
+
"|<EXTRA_TOKENS_235>|",
|
| 2667 |
+
"|<EXTRA_TOKENS_236>|",
|
| 2668 |
+
"|<EXTRA_TOKENS_237>|",
|
| 2669 |
+
"|<EXTRA_TOKENS_238>|",
|
| 2670 |
+
"|<EXTRA_TOKENS_239>|",
|
| 2671 |
+
"|<EXTRA_TOKENS_240>|",
|
| 2672 |
+
"|<EXTRA_TOKENS_241>|",
|
| 2673 |
+
"|<EXTRA_TOKENS_242>|",
|
| 2674 |
+
"|<EXTRA_TOKENS_243>|",
|
| 2675 |
+
"|<EXTRA_TOKENS_244>|",
|
| 2676 |
+
"|<EXTRA_TOKENS_245>|",
|
| 2677 |
+
"|<EXTRA_TOKENS_246>|",
|
| 2678 |
+
"|<EXTRA_TOKENS_247>|",
|
| 2679 |
+
"|<EXTRA_TOKENS_248>|",
|
| 2680 |
+
"|<EXTRA_TOKENS_249>|",
|
| 2681 |
+
"|<EXTRA_TOKENS_250>|",
|
| 2682 |
+
"|<EXTRA_TOKENS_251>|",
|
| 2683 |
+
"|<EXTRA_TOKENS_252>|",
|
| 2684 |
+
"|<EXTRA_TOKENS_253>|",
|
| 2685 |
+
"|<EXTRA_TOKENS_254>|",
|
| 2686 |
+
"|<EXTRA_TOKENS_255>|",
|
| 2687 |
+
"|<EXTRA_TOKENS_256>|",
|
| 2688 |
+
"|<EXTRA_TOKENS_257>|",
|
| 2689 |
+
"|<EXTRA_TOKENS_258>|",
|
| 2690 |
+
"|<EXTRA_TOKENS_259>|",
|
| 2691 |
+
"|<EXTRA_TOKENS_260>|",
|
| 2692 |
+
"|<EXTRA_TOKENS_261>|",
|
| 2693 |
+
"|<EXTRA_TOKENS_262>|",
|
| 2694 |
+
"|<EXTRA_TOKENS_263>|",
|
| 2695 |
+
"|<EXTRA_TOKENS_264>|",
|
| 2696 |
+
"|<EXTRA_TOKENS_265>|",
|
| 2697 |
+
"|<EXTRA_TOKENS_266>|",
|
| 2698 |
+
"<im_start>",
|
| 2699 |
+
"<im_end>",
|
| 2700 |
+
"<im_patch>",
|
| 2701 |
+
"<im_col>",
|
| 2702 |
+
"<low_res_im_start>",
|
| 2703 |
+
"<|image|>",
|
| 2704 |
+
"<im_low>",
|
| 2705 |
+
"<frame_start>",
|
| 2706 |
+
"<frame_end>",
|
| 2707 |
+
"<|video|>"
|
| 2708 |
+
],
|
| 2709 |
+
"auto_map": {
|
| 2710 |
+
"AutoProcessor": "processing_molmo2.Molmo2Processor"
|
| 2711 |
+
},
|
| 2712 |
+
"bos_token": "<|im_end|>",
|
| 2713 |
+
"clean_up_tokenization_spaces": false,
|
| 2714 |
+
"eos_token": "<|im_end|>",
|
| 2715 |
+
"errors": "replace",
|
| 2716 |
+
"extra_special_tokens": {},
|
| 2717 |
+
"model_max_length": 131072,
|
| 2718 |
+
"pad_token": "<|endoftext|>",
|
| 2719 |
+
"processor_class": "Molmo2Processor",
|
| 2720 |
+
"split_special_tokens": false,
|
| 2721 |
+
"tokenizer_class": "Qwen2Tokenizer",
|
| 2722 |
+
"unk_token": null,
|
| 2723 |
+
"eos_token_id": 151645,
|
| 2724 |
+
"bos_token_id": 151645,
|
| 2725 |
+
"pad_token_id": 151643
|
| 2726 |
+
}
|
train/rank0.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f9dcc4e07716c7ccf2dd52a03ec28ba04aecc3ca6bad4d29a05fdf531f24d01b
|
| 3 |
+
size 17877
|
train/rank1.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a8fcda81d62ab1b4bedddf90161f3a21c9d9d771ed5251450117b2a498e2168d
|
| 3 |
+
size 17877
|
train/rank10.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd54a1b6c04f3a5ff476d0814048de3afe45189d64aaf0cd48f2abb1098d4471
|
| 3 |
+
size 17885
|
train/rank11.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f9846ea2a9267545038d490300087334a67203c91ccc620a46ffcbb0252ae345
|
| 3 |
+
size 17885
|
train/rank12.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62643ae3166ca300ab7a25efe55a18a43c2c372ad5169b7a68ea45736bd6d539
|
| 3 |
+
size 17885
|
train/rank13.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e00eaa7e08b413805e88db583250da0ee0e46e5418879c5cfb0b52adda60e644
|
| 3 |
+
size 17885
|
train/rank14.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3ef140976efb7f4112304a3a006009c07b0d2b41ef67123edff23d5c8a87f6e0
|
| 3 |
+
size 17885
|
train/rank15.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2d3a77701c27c950181ab430fed54562be4919a47cd98176247a65cff6ff477c
|
| 3 |
+
size 17885
|
train/rank16.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:096e3924afb1f6e9c56903b187fc56f56cd2181acb8b08f5cd29ad254e39cc84
|
| 3 |
+
size 17885
|
train/rank17.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d670498b95f098c9557ae78b5f4e6e39913f087adea6a1be59d202c6ccae053d
|
| 3 |
+
size 17885
|
train/rank18.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ff37d51108ce6851dac0d46bca854af5f5ede648809c4783b02ba251fa0d73a8
|
| 3 |
+
size 17885
|
train/rank19.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1f4bf250a23bddc41105b25d808cb92c1f1178944c4c0704c427fd6dca4ba6d
|
| 3 |
+
size 17885
|
train/rank2.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7e25077ec4da0b9279ca08770814a789dc1f04d9900bf1e0f2c2f8f34e8b341
|
| 3 |
+
size 17877
|
train/rank20.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:97474bc5b2ef73df5e3b001832f992d1c792bc8668aea13fb2f82fb1669d5ed4
|
| 3 |
+
size 17885
|
train/rank21.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6c9ee4f5d41fd4ea38f24fad88d2b524b85a19eef141c55608c9f9edc560b109
|
| 3 |
+
size 17885
|
train/rank22.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0723c1595590d926cb5f0a627ce353a792f5e6b8b49236ce35d4349d00feb2e4
|
| 3 |
+
size 17885
|
train/rank23.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:101cb5271d7b56c9e54c5eb6032827a516d3479473bb4785bb4d6eccfd7394e0
|
| 3 |
+
size 17885
|
train/rank24.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5db5ff25d6606e8902b4269df33d07e2c52e6571545f8edcc8adffc877c012ae
|
| 3 |
+
size 17885
|
train/rank25.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f073249240ab66d1741b7b5529f36ffaeab844106211d30bbdb37f2baf7864b
|
| 3 |
+
size 17885
|
train/rank26.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b48932e534a8962846c1c34e9e13eb5f7877d95ee0d46e168a5e0ed89174b4c4
|
| 3 |
+
size 17885
|
train/rank27.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:56718b5cf614fd133a30492f9bbc381f0bdf4bf437e8233fc2460ff032a9b4df
|
| 3 |
+
size 17885
|
train/rank28.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2288df20edae6dd5dc0442f26dc986f03b83b5cad42123a51cafb4b243441a3b
|
| 3 |
+
size 17885
|
train/rank29.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:397530577ac9643eb2ec47167f75484f6dd6790bf20589ded4a5c18878133283
|
| 3 |
+
size 17885
|
train/rank3.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ba8014748f2ad6d684d0f63b1e85e18a4574721ef8a36e7a1950377e76d800a8
|
| 3 |
+
size 17877
|
train/rank30.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:443ae57071c2a5eece0868494c2aad7b4b3ee290688bcf9b4a720fd7d87a132a
|
| 3 |
+
size 17885
|
train/rank31.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1c6b559828dc6f039b5ce776bbbd2828c14eb13a29c8be1cd4088211e04b5695
|
| 3 |
+
size 17885
|
train/rank32.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81a5262bde1e0662c78b76cc75c77832e115f00bdb8242b258ed69d146b4469a
|
| 3 |
+
size 17885
|