Upload FlashJudge artifacts from run_20260505_225612

flash_judge_checkpoint.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d022887bca2e346a8eb836c1484caf22a55b4853744ad12c61e6a06a817927f
+size 1543637939

training_config.json

@@ -0,0 +1,83 @@
+{
+  "model_id": "google/siglip2-base-patch16-naflex",
+  "dataset": "apartments",
+  "apartments_jsonl": "data/apartments/question-image-dataset_labeled_expanded.jsonl",
+  "apartments_images_dir": "data/apartments/images",
+  "fmnist_root": "data",
+  "output_root": "models",
+  "run_name": null,
+  "batch_size": 20,
+  "gradient_accumulation_steps": 1,
+  "stack_size": 1,
+  "epochs": 2,
+  "max_steps": null,
+  "eval_interval": 1,
+  "eval_batches": 2,
+  "checkpoint_interval": 1000,
+  "checkpoint_mode": "max_val_acc",
+  "val_ratio": 0.1,
+  "seed": 42,
+  "apartments_train_group_limit": null,
+  "apartments_train_question_limit": null,
+  "apartments_question_split_mode": "disjoint",
+  "apartments_hard_negative_strategy": "within_group_overlap",
+  "apartments_hard_negative_topk": 0,
+  "apartments_hard_negative_oversample_factor": 0,
+  "apartments_augment_images": true,
+  "apartments_augment_add_prob": 0.35,
+  "apartments_augment_max_added_images": 3,
+  "apartments_augment_shuffle_prob": 0.5,
+  "apartments_augment_repeat_prob": 0.25,
+  "apartments_augment_repeat_max_images": 2,
+  "apartments_augment_rotate_prob": 0.5,
+  "apartments_augment_rotate_degrees": 5.0,
+  "apartments_augment_color_prob": 0.8,
+  "apartments_augment_brightness_delta": 0.1,
+  "apartments_augment_saturation_delta": 0.1,
+  "num_workers_train": 4,
+  "num_workers_val": 2,
+  "text_max_length": 64,
+  "image_max_num_patches": null,
+  "lr_backbone": 1e-6,
+  "lr_head": 0.0002,
+  "weight_decay": 0.01,
+  "freeze_backbone": false,
+  "freeze_backbone_epochs": 0,
+  "freeze_backbone_steps": null,
+  "backbone_trainable_scope": "full",
+  "attention_heads": 2,
+  "attention_dropout": 0.3,
+  "image_self_attention_layers": 0,
+  "image_dropout": 0.15,
+  "normalize_embeddings": true,
+  "set_context_mode": "mean",
+  "aggregation_mode": "attention",
+  "head_feature_mode": "matching",
+  "late_interaction_topk": 8,
+  "head_hidden_dim": 768,
+  "head_num_layers": 2,
+  "head_dropout": 0.2,
+  "head_activation": "swiglu",
+  "label_smoothing": 0.05,
+  "alignment_loss_weight": 0.05,
+  "alignment_loss_logit_scale": 18.0,
+  "alignment_loss_hard_negative_topk": 1,
+  "compile_model": false,
+  "compile_fallback_to_eager": true,
+  "backbone_torch_dtype": "auto",
+  "attn_implementation": null,
+  "resume_path": null,
+  "min_free_space_gb": 1.0,
+  "use_wandb": true,
+  "wandb_project": "flash-judge",
+  "wandb_entity": null,
+  "wandb_run_name": null,
+  "save_artifacts": true,
+  "push_to_hub": true,
+  "hub_org": "FuncAI",
+  "hub_repo_name": "FlashJudge7",
+  "hub_private": false,
+  "hub_token": null,
+  "device": null,
+  "use_amp": true
+}

training_logs.csv

@@ -0,0 +1,3 @@
+epoch,train_loss,train_acc,val_loss,val_acc,max_val_acc
+1,0.3626155336048879,0.8760223448383815,0.36394935452234245,0.8483606597530793,0.8483606597530793
+2,0.20038580257494265,0.969442161696799,0.39457498462947055,0.8397541029806264,0.8483606597530793