Upload FlashJudge artifacts from run_20260307_174019
Browse files- flash_judge_checkpoint.pth +3 -0
- training_config.json +48 -0
- training_logs.csv +0 -0
flash_judge_checkpoint.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b7e3f9e7760ecc60accb43f77dbb51ed4d356d32537f39442cf7de9fc9621373
|
| 3 |
+
size 1513528529
|
training_config.json
ADDED
|
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_id": "google/siglip2-base-patch16-224",
|
| 3 |
+
"dataset": "apartments",
|
| 4 |
+
"apartments_jsonl": "data/apartments/question-image-dataset_labeled.jsonl",
|
| 5 |
+
"apartments_images_dir": "data/apartments/images",
|
| 6 |
+
"fmnist_root": "data",
|
| 7 |
+
"output_root": "models",
|
| 8 |
+
"run_name": null,
|
| 9 |
+
"batch_size": 8,
|
| 10 |
+
"stack_size": 1,
|
| 11 |
+
"max_steps": 5000,
|
| 12 |
+
"eval_interval": 10,
|
| 13 |
+
"eval_batches": 2,
|
| 14 |
+
"checkpoint_interval": 1000,
|
| 15 |
+
"checkpoint_mode": "max_val_acc",
|
| 16 |
+
"val_ratio": 0.1,
|
| 17 |
+
"seed": 42,
|
| 18 |
+
"num_workers_train": 4,
|
| 19 |
+
"num_workers_val": 2,
|
| 20 |
+
"lr_backbone": 2e-6,
|
| 21 |
+
"lr_head": 0.0003,
|
| 22 |
+
"weight_decay": 0.0,
|
| 23 |
+
"freeze_backbone": false,
|
| 24 |
+
"attention_heads": 2,
|
| 25 |
+
"attention_dropout": 0.3,
|
| 26 |
+
"image_self_attention_layers": 0,
|
| 27 |
+
"head_hidden_dim": 512,
|
| 28 |
+
"head_num_layers": 1,
|
| 29 |
+
"head_dropout": 0.3,
|
| 30 |
+
"head_activation": "swiglu",
|
| 31 |
+
"label_smoothing": 0.05,
|
| 32 |
+
"compile_model": false,
|
| 33 |
+
"compile_fallback_to_eager": true,
|
| 34 |
+
"resume_path": null,
|
| 35 |
+
"min_free_space_gb": 1.0,
|
| 36 |
+
"use_wandb": true,
|
| 37 |
+
"wandb_project": "flash-judge",
|
| 38 |
+
"wandb_entity": null,
|
| 39 |
+
"wandb_run_name": null,
|
| 40 |
+
"save_artifacts": true,
|
| 41 |
+
"push_to_hub": true,
|
| 42 |
+
"hub_org": "FuncAI",
|
| 43 |
+
"hub_repo_name": "FlashJudge",
|
| 44 |
+
"hub_private": false,
|
| 45 |
+
"hub_token": null,
|
| 46 |
+
"device": null,
|
| 47 |
+
"use_amp": true
|
| 48 |
+
}
|
training_logs.csv
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|