Upload folder using huggingface_hub
Browse files- checkpoint-1548/config.json +112 -0
- checkpoint-1548/model.safetensors +3 -0
- checkpoint-1548/optimizer.pt +3 -0
- checkpoint-1548/preprocessor_config.json +24 -0
- checkpoint-1548/rng_state.pth +3 -0
- checkpoint-1548/scheduler.pt +3 -0
- checkpoint-1548/trainer_state.json +75 -0
- checkpoint-1548/training_args.bin +3 -0
- checkpoint-774/config.json +112 -0
- checkpoint-774/model.safetensors +3 -0
- checkpoint-774/optimizer.pt +3 -0
- checkpoint-774/preprocessor_config.json +24 -0
- checkpoint-774/rng_state.pth +3 -0
- checkpoint-774/scheduler.pt +3 -0
- checkpoint-774/trainer_state.json +51 -0
- checkpoint-774/training_args.bin +3 -0
- config.json +112 -0
- model.safetensors +3 -0
- preprocessor_config.json +24 -0
- training_args.bin +3 -0
checkpoint-1548/config.json
ADDED
|
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"SiglipForImageClassification"
|
| 4 |
+
],
|
| 5 |
+
"id2label": {
|
| 6 |
+
"0": "parking lot",
|
| 7 |
+
"1": "avenue",
|
| 8 |
+
"2": "highway",
|
| 9 |
+
"3": "bridge",
|
| 10 |
+
"4": "marina",
|
| 11 |
+
"5": "crossroads",
|
| 12 |
+
"6": "airport runway",
|
| 13 |
+
"7": "pipeline",
|
| 14 |
+
"8": "town",
|
| 15 |
+
"9": "airplane",
|
| 16 |
+
"10": "forest",
|
| 17 |
+
"11": "mangrove",
|
| 18 |
+
"12": "artificial grassland",
|
| 19 |
+
"13": "river protection forest",
|
| 20 |
+
"14": "shrubwood",
|
| 21 |
+
"15": "sapling",
|
| 22 |
+
"16": "sparse forest",
|
| 23 |
+
"17": "lakeshore",
|
| 24 |
+
"18": "river",
|
| 25 |
+
"19": "stream",
|
| 26 |
+
"20": "coastline",
|
| 27 |
+
"21": "hirst",
|
| 28 |
+
"22": "dam",
|
| 29 |
+
"23": "sea",
|
| 30 |
+
"24": "snow mountain",
|
| 31 |
+
"25": "sandbeach",
|
| 32 |
+
"26": "mountain",
|
| 33 |
+
"27": "desert",
|
| 34 |
+
"28": "dry farm",
|
| 35 |
+
"29": "green farmland",
|
| 36 |
+
"30": "bare land",
|
| 37 |
+
"31": "city building",
|
| 38 |
+
"32": "residents",
|
| 39 |
+
"33": "container",
|
| 40 |
+
"34": "storage room"
|
| 41 |
+
},
|
| 42 |
+
"initializer_factor": 1.0,
|
| 43 |
+
"label2id": {
|
| 44 |
+
"airplane": 9,
|
| 45 |
+
"airport runway": 6,
|
| 46 |
+
"artificial grassland": 12,
|
| 47 |
+
"avenue": 1,
|
| 48 |
+
"bare land": 30,
|
| 49 |
+
"bridge": 3,
|
| 50 |
+
"city building": 31,
|
| 51 |
+
"coastline": 20,
|
| 52 |
+
"container": 33,
|
| 53 |
+
"crossroads": 5,
|
| 54 |
+
"dam": 22,
|
| 55 |
+
"desert": 27,
|
| 56 |
+
"dry farm": 28,
|
| 57 |
+
"forest": 10,
|
| 58 |
+
"green farmland": 29,
|
| 59 |
+
"highway": 2,
|
| 60 |
+
"hirst": 21,
|
| 61 |
+
"lakeshore": 17,
|
| 62 |
+
"mangrove": 11,
|
| 63 |
+
"marina": 4,
|
| 64 |
+
"mountain": 26,
|
| 65 |
+
"parking lot": 0,
|
| 66 |
+
"pipeline": 7,
|
| 67 |
+
"residents": 32,
|
| 68 |
+
"river": 18,
|
| 69 |
+
"river protection forest": 13,
|
| 70 |
+
"sandbeach": 25,
|
| 71 |
+
"sapling": 15,
|
| 72 |
+
"sea": 23,
|
| 73 |
+
"shrubwood": 14,
|
| 74 |
+
"snow mountain": 24,
|
| 75 |
+
"sparse forest": 16,
|
| 76 |
+
"storage room": 34,
|
| 77 |
+
"stream": 19,
|
| 78 |
+
"town": 8
|
| 79 |
+
},
|
| 80 |
+
"model_type": "siglip",
|
| 81 |
+
"problem_type": "single_label_classification",
|
| 82 |
+
"text_config": {
|
| 83 |
+
"attention_dropout": 0.0,
|
| 84 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 85 |
+
"hidden_size": 768,
|
| 86 |
+
"intermediate_size": 3072,
|
| 87 |
+
"layer_norm_eps": 1e-06,
|
| 88 |
+
"max_position_embeddings": 64,
|
| 89 |
+
"model_type": "siglip_text_model",
|
| 90 |
+
"num_attention_heads": 12,
|
| 91 |
+
"num_hidden_layers": 12,
|
| 92 |
+
"projection_size": 768,
|
| 93 |
+
"torch_dtype": "float32",
|
| 94 |
+
"vocab_size": 256000
|
| 95 |
+
},
|
| 96 |
+
"torch_dtype": "float32",
|
| 97 |
+
"transformers_version": "4.50.0",
|
| 98 |
+
"vision_config": {
|
| 99 |
+
"attention_dropout": 0.0,
|
| 100 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 101 |
+
"hidden_size": 768,
|
| 102 |
+
"image_size": 224,
|
| 103 |
+
"intermediate_size": 3072,
|
| 104 |
+
"layer_norm_eps": 1e-06,
|
| 105 |
+
"model_type": "siglip_vision_model",
|
| 106 |
+
"num_attention_heads": 12,
|
| 107 |
+
"num_channels": 3,
|
| 108 |
+
"num_hidden_layers": 12,
|
| 109 |
+
"patch_size": 16,
|
| 110 |
+
"torch_dtype": "float32"
|
| 111 |
+
}
|
| 112 |
+
}
|
checkpoint-1548/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e396dbc84e31c71200f42dd0ba85793307bc33ef3ef0d03c42a8f07724264853
|
| 3 |
+
size 371669508
|
checkpoint-1548/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9d501155e33d77e4995e96989f405bb79050c05c4b136f3c58329a30d6e70581
|
| 3 |
+
size 686758778
|
checkpoint-1548/preprocessor_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_rgb": null,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"image_mean": [
|
| 7 |
+
0.5,
|
| 8 |
+
0.5,
|
| 9 |
+
0.5
|
| 10 |
+
],
|
| 11 |
+
"image_processor_type": "SiglipImageProcessor",
|
| 12 |
+
"image_std": [
|
| 13 |
+
0.5,
|
| 14 |
+
0.5,
|
| 15 |
+
0.5
|
| 16 |
+
],
|
| 17 |
+
"processor_class": "SiglipProcessor",
|
| 18 |
+
"resample": 2,
|
| 19 |
+
"rescale_factor": 0.00392156862745098,
|
| 20 |
+
"size": {
|
| 21 |
+
"height": 224,
|
| 22 |
+
"width": 224
|
| 23 |
+
}
|
| 24 |
+
}
|
checkpoint-1548/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a127e431fd8b5f93a3bf02626092859ebe2bfdd882e2b2e537bbd13addc9cec2
|
| 3 |
+
size 14244
|
checkpoint-1548/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb050df0950ff8e921910b930efd06056f7447f89133e0edcb782e0a9335c2d8
|
| 3 |
+
size 1064
|
checkpoint-1548/trainer_state.json
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 1548,
|
| 3 |
+
"best_metric": 0.023793132975697517,
|
| 4 |
+
"best_model_checkpoint": "siglip2-finetune-full/checkpoint-1548",
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1548,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.6459948320413437,
|
| 14 |
+
"grad_norm": 10.552935600280762,
|
| 15 |
+
"learning_rate": 0.00013991989319092122,
|
| 16 |
+
"loss": 0.6623,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 1.0,
|
| 21 |
+
"eval_accuracy": 0.9349820180223866,
|
| 22 |
+
"eval_loss": 0.21081215143203735,
|
| 23 |
+
"eval_model_preparation_time": 0.0023,
|
| 24 |
+
"eval_runtime": 329.2887,
|
| 25 |
+
"eval_samples_per_second": 75.153,
|
| 26 |
+
"eval_steps_per_second": 9.396,
|
| 27 |
+
"step": 774
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 1.2919896640826873,
|
| 31 |
+
"grad_norm": 2.3672032356262207,
|
| 32 |
+
"learning_rate": 7.316421895861149e-05,
|
| 33 |
+
"loss": 0.2013,
|
| 34 |
+
"step": 1000
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 1.937984496124031,
|
| 38 |
+
"grad_norm": 0.24311719834804535,
|
| 39 |
+
"learning_rate": 6.408544726301736e-06,
|
| 40 |
+
"loss": 0.0687,
|
| 41 |
+
"step": 1500
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 2.0,
|
| 45 |
+
"eval_accuracy": 0.9919182123085627,
|
| 46 |
+
"eval_loss": 0.023793132975697517,
|
| 47 |
+
"eval_model_preparation_time": 0.0023,
|
| 48 |
+
"eval_runtime": 328.7926,
|
| 49 |
+
"eval_samples_per_second": 75.266,
|
| 50 |
+
"eval_steps_per_second": 9.41,
|
| 51 |
+
"step": 1548
|
| 52 |
+
}
|
| 53 |
+
],
|
| 54 |
+
"logging_steps": 500,
|
| 55 |
+
"max_steps": 1548,
|
| 56 |
+
"num_input_tokens_seen": 0,
|
| 57 |
+
"num_train_epochs": 2,
|
| 58 |
+
"save_steps": 500,
|
| 59 |
+
"stateful_callbacks": {
|
| 60 |
+
"TrainerControl": {
|
| 61 |
+
"args": {
|
| 62 |
+
"should_epoch_stop": false,
|
| 63 |
+
"should_evaluate": false,
|
| 64 |
+
"should_log": false,
|
| 65 |
+
"should_save": true,
|
| 66 |
+
"should_training_stop": true
|
| 67 |
+
},
|
| 68 |
+
"attributes": {}
|
| 69 |
+
}
|
| 70 |
+
},
|
| 71 |
+
"total_flos": 4.146528897531482e+18,
|
| 72 |
+
"train_batch_size": 32,
|
| 73 |
+
"trial_name": null,
|
| 74 |
+
"trial_params": null
|
| 75 |
+
}
|
checkpoint-1548/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12dd34e788f5eb534c4b6ac8f57ace245fefa2a536daa2fe035c88d0385d96e8
|
| 3 |
+
size 5304
|
checkpoint-774/config.json
ADDED
|
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"SiglipForImageClassification"
|
| 4 |
+
],
|
| 5 |
+
"id2label": {
|
| 6 |
+
"0": "parking lot",
|
| 7 |
+
"1": "avenue",
|
| 8 |
+
"2": "highway",
|
| 9 |
+
"3": "bridge",
|
| 10 |
+
"4": "marina",
|
| 11 |
+
"5": "crossroads",
|
| 12 |
+
"6": "airport runway",
|
| 13 |
+
"7": "pipeline",
|
| 14 |
+
"8": "town",
|
| 15 |
+
"9": "airplane",
|
| 16 |
+
"10": "forest",
|
| 17 |
+
"11": "mangrove",
|
| 18 |
+
"12": "artificial grassland",
|
| 19 |
+
"13": "river protection forest",
|
| 20 |
+
"14": "shrubwood",
|
| 21 |
+
"15": "sapling",
|
| 22 |
+
"16": "sparse forest",
|
| 23 |
+
"17": "lakeshore",
|
| 24 |
+
"18": "river",
|
| 25 |
+
"19": "stream",
|
| 26 |
+
"20": "coastline",
|
| 27 |
+
"21": "hirst",
|
| 28 |
+
"22": "dam",
|
| 29 |
+
"23": "sea",
|
| 30 |
+
"24": "snow mountain",
|
| 31 |
+
"25": "sandbeach",
|
| 32 |
+
"26": "mountain",
|
| 33 |
+
"27": "desert",
|
| 34 |
+
"28": "dry farm",
|
| 35 |
+
"29": "green farmland",
|
| 36 |
+
"30": "bare land",
|
| 37 |
+
"31": "city building",
|
| 38 |
+
"32": "residents",
|
| 39 |
+
"33": "container",
|
| 40 |
+
"34": "storage room"
|
| 41 |
+
},
|
| 42 |
+
"initializer_factor": 1.0,
|
| 43 |
+
"label2id": {
|
| 44 |
+
"airplane": 9,
|
| 45 |
+
"airport runway": 6,
|
| 46 |
+
"artificial grassland": 12,
|
| 47 |
+
"avenue": 1,
|
| 48 |
+
"bare land": 30,
|
| 49 |
+
"bridge": 3,
|
| 50 |
+
"city building": 31,
|
| 51 |
+
"coastline": 20,
|
| 52 |
+
"container": 33,
|
| 53 |
+
"crossroads": 5,
|
| 54 |
+
"dam": 22,
|
| 55 |
+
"desert": 27,
|
| 56 |
+
"dry farm": 28,
|
| 57 |
+
"forest": 10,
|
| 58 |
+
"green farmland": 29,
|
| 59 |
+
"highway": 2,
|
| 60 |
+
"hirst": 21,
|
| 61 |
+
"lakeshore": 17,
|
| 62 |
+
"mangrove": 11,
|
| 63 |
+
"marina": 4,
|
| 64 |
+
"mountain": 26,
|
| 65 |
+
"parking lot": 0,
|
| 66 |
+
"pipeline": 7,
|
| 67 |
+
"residents": 32,
|
| 68 |
+
"river": 18,
|
| 69 |
+
"river protection forest": 13,
|
| 70 |
+
"sandbeach": 25,
|
| 71 |
+
"sapling": 15,
|
| 72 |
+
"sea": 23,
|
| 73 |
+
"shrubwood": 14,
|
| 74 |
+
"snow mountain": 24,
|
| 75 |
+
"sparse forest": 16,
|
| 76 |
+
"storage room": 34,
|
| 77 |
+
"stream": 19,
|
| 78 |
+
"town": 8
|
| 79 |
+
},
|
| 80 |
+
"model_type": "siglip",
|
| 81 |
+
"problem_type": "single_label_classification",
|
| 82 |
+
"text_config": {
|
| 83 |
+
"attention_dropout": 0.0,
|
| 84 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 85 |
+
"hidden_size": 768,
|
| 86 |
+
"intermediate_size": 3072,
|
| 87 |
+
"layer_norm_eps": 1e-06,
|
| 88 |
+
"max_position_embeddings": 64,
|
| 89 |
+
"model_type": "siglip_text_model",
|
| 90 |
+
"num_attention_heads": 12,
|
| 91 |
+
"num_hidden_layers": 12,
|
| 92 |
+
"projection_size": 768,
|
| 93 |
+
"torch_dtype": "float32",
|
| 94 |
+
"vocab_size": 256000
|
| 95 |
+
},
|
| 96 |
+
"torch_dtype": "float32",
|
| 97 |
+
"transformers_version": "4.50.0",
|
| 98 |
+
"vision_config": {
|
| 99 |
+
"attention_dropout": 0.0,
|
| 100 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 101 |
+
"hidden_size": 768,
|
| 102 |
+
"image_size": 224,
|
| 103 |
+
"intermediate_size": 3072,
|
| 104 |
+
"layer_norm_eps": 1e-06,
|
| 105 |
+
"model_type": "siglip_vision_model",
|
| 106 |
+
"num_attention_heads": 12,
|
| 107 |
+
"num_channels": 3,
|
| 108 |
+
"num_hidden_layers": 12,
|
| 109 |
+
"patch_size": 16,
|
| 110 |
+
"torch_dtype": "float32"
|
| 111 |
+
}
|
| 112 |
+
}
|
checkpoint-774/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ecb01b3715732069c8e59056c7c4104c25a601462ecd7dd1d23c604300ef7ec8
|
| 3 |
+
size 371669508
|
checkpoint-774/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a87476ba95c0064fb2e21c9e3bac167c22b77ee637c9fd6d936448c03c35b998
|
| 3 |
+
size 686758778
|
checkpoint-774/preprocessor_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_rgb": null,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"image_mean": [
|
| 7 |
+
0.5,
|
| 8 |
+
0.5,
|
| 9 |
+
0.5
|
| 10 |
+
],
|
| 11 |
+
"image_processor_type": "SiglipImageProcessor",
|
| 12 |
+
"image_std": [
|
| 13 |
+
0.5,
|
| 14 |
+
0.5,
|
| 15 |
+
0.5
|
| 16 |
+
],
|
| 17 |
+
"processor_class": "SiglipProcessor",
|
| 18 |
+
"resample": 2,
|
| 19 |
+
"rescale_factor": 0.00392156862745098,
|
| 20 |
+
"size": {
|
| 21 |
+
"height": 224,
|
| 22 |
+
"width": 224
|
| 23 |
+
}
|
| 24 |
+
}
|
checkpoint-774/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c77f7ded6171386b451b70d83d116dd0614aba90de316cb058986d6802fc9826
|
| 3 |
+
size 14244
|
checkpoint-774/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:db8b0e48a825de1ef288d11a2c2c7eeb56cc8010ca0357de1b98d6e304b36c63
|
| 3 |
+
size 1064
|
checkpoint-774/trainer_state.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 774,
|
| 3 |
+
"best_metric": 0.21081215143203735,
|
| 4 |
+
"best_model_checkpoint": "siglip2-finetune-full/checkpoint-774",
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 774,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.6459948320413437,
|
| 14 |
+
"grad_norm": 10.552935600280762,
|
| 15 |
+
"learning_rate": 0.00013991989319092122,
|
| 16 |
+
"loss": 0.6623,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 1.0,
|
| 21 |
+
"eval_accuracy": 0.9349820180223866,
|
| 22 |
+
"eval_loss": 0.21081215143203735,
|
| 23 |
+
"eval_model_preparation_time": 0.0023,
|
| 24 |
+
"eval_runtime": 329.2887,
|
| 25 |
+
"eval_samples_per_second": 75.153,
|
| 26 |
+
"eval_steps_per_second": 9.396,
|
| 27 |
+
"step": 774
|
| 28 |
+
}
|
| 29 |
+
],
|
| 30 |
+
"logging_steps": 500,
|
| 31 |
+
"max_steps": 1548,
|
| 32 |
+
"num_input_tokens_seen": 0,
|
| 33 |
+
"num_train_epochs": 2,
|
| 34 |
+
"save_steps": 500,
|
| 35 |
+
"stateful_callbacks": {
|
| 36 |
+
"TrainerControl": {
|
| 37 |
+
"args": {
|
| 38 |
+
"should_epoch_stop": false,
|
| 39 |
+
"should_evaluate": false,
|
| 40 |
+
"should_log": false,
|
| 41 |
+
"should_save": true,
|
| 42 |
+
"should_training_stop": false
|
| 43 |
+
},
|
| 44 |
+
"attributes": {}
|
| 45 |
+
}
|
| 46 |
+
},
|
| 47 |
+
"total_flos": 2.073264448765741e+18,
|
| 48 |
+
"train_batch_size": 32,
|
| 49 |
+
"trial_name": null,
|
| 50 |
+
"trial_params": null
|
| 51 |
+
}
|
checkpoint-774/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12dd34e788f5eb534c4b6ac8f57ace245fefa2a536daa2fe035c88d0385d96e8
|
| 3 |
+
size 5304
|
config.json
ADDED
|
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"SiglipForImageClassification"
|
| 4 |
+
],
|
| 5 |
+
"id2label": {
|
| 6 |
+
"0": "parking lot",
|
| 7 |
+
"1": "avenue",
|
| 8 |
+
"2": "highway",
|
| 9 |
+
"3": "bridge",
|
| 10 |
+
"4": "marina",
|
| 11 |
+
"5": "crossroads",
|
| 12 |
+
"6": "airport runway",
|
| 13 |
+
"7": "pipeline",
|
| 14 |
+
"8": "town",
|
| 15 |
+
"9": "airplane",
|
| 16 |
+
"10": "forest",
|
| 17 |
+
"11": "mangrove",
|
| 18 |
+
"12": "artificial grassland",
|
| 19 |
+
"13": "river protection forest",
|
| 20 |
+
"14": "shrubwood",
|
| 21 |
+
"15": "sapling",
|
| 22 |
+
"16": "sparse forest",
|
| 23 |
+
"17": "lakeshore",
|
| 24 |
+
"18": "river",
|
| 25 |
+
"19": "stream",
|
| 26 |
+
"20": "coastline",
|
| 27 |
+
"21": "hirst",
|
| 28 |
+
"22": "dam",
|
| 29 |
+
"23": "sea",
|
| 30 |
+
"24": "snow mountain",
|
| 31 |
+
"25": "sandbeach",
|
| 32 |
+
"26": "mountain",
|
| 33 |
+
"27": "desert",
|
| 34 |
+
"28": "dry farm",
|
| 35 |
+
"29": "green farmland",
|
| 36 |
+
"30": "bare land",
|
| 37 |
+
"31": "city building",
|
| 38 |
+
"32": "residents",
|
| 39 |
+
"33": "container",
|
| 40 |
+
"34": "storage room"
|
| 41 |
+
},
|
| 42 |
+
"initializer_factor": 1.0,
|
| 43 |
+
"label2id": {
|
| 44 |
+
"airplane": 9,
|
| 45 |
+
"airport runway": 6,
|
| 46 |
+
"artificial grassland": 12,
|
| 47 |
+
"avenue": 1,
|
| 48 |
+
"bare land": 30,
|
| 49 |
+
"bridge": 3,
|
| 50 |
+
"city building": 31,
|
| 51 |
+
"coastline": 20,
|
| 52 |
+
"container": 33,
|
| 53 |
+
"crossroads": 5,
|
| 54 |
+
"dam": 22,
|
| 55 |
+
"desert": 27,
|
| 56 |
+
"dry farm": 28,
|
| 57 |
+
"forest": 10,
|
| 58 |
+
"green farmland": 29,
|
| 59 |
+
"highway": 2,
|
| 60 |
+
"hirst": 21,
|
| 61 |
+
"lakeshore": 17,
|
| 62 |
+
"mangrove": 11,
|
| 63 |
+
"marina": 4,
|
| 64 |
+
"mountain": 26,
|
| 65 |
+
"parking lot": 0,
|
| 66 |
+
"pipeline": 7,
|
| 67 |
+
"residents": 32,
|
| 68 |
+
"river": 18,
|
| 69 |
+
"river protection forest": 13,
|
| 70 |
+
"sandbeach": 25,
|
| 71 |
+
"sapling": 15,
|
| 72 |
+
"sea": 23,
|
| 73 |
+
"shrubwood": 14,
|
| 74 |
+
"snow mountain": 24,
|
| 75 |
+
"sparse forest": 16,
|
| 76 |
+
"storage room": 34,
|
| 77 |
+
"stream": 19,
|
| 78 |
+
"town": 8
|
| 79 |
+
},
|
| 80 |
+
"model_type": "siglip",
|
| 81 |
+
"problem_type": "single_label_classification",
|
| 82 |
+
"text_config": {
|
| 83 |
+
"attention_dropout": 0.0,
|
| 84 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 85 |
+
"hidden_size": 768,
|
| 86 |
+
"intermediate_size": 3072,
|
| 87 |
+
"layer_norm_eps": 1e-06,
|
| 88 |
+
"max_position_embeddings": 64,
|
| 89 |
+
"model_type": "siglip_text_model",
|
| 90 |
+
"num_attention_heads": 12,
|
| 91 |
+
"num_hidden_layers": 12,
|
| 92 |
+
"projection_size": 768,
|
| 93 |
+
"torch_dtype": "float32",
|
| 94 |
+
"vocab_size": 256000
|
| 95 |
+
},
|
| 96 |
+
"torch_dtype": "float32",
|
| 97 |
+
"transformers_version": "4.50.0",
|
| 98 |
+
"vision_config": {
|
| 99 |
+
"attention_dropout": 0.0,
|
| 100 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 101 |
+
"hidden_size": 768,
|
| 102 |
+
"image_size": 224,
|
| 103 |
+
"intermediate_size": 3072,
|
| 104 |
+
"layer_norm_eps": 1e-06,
|
| 105 |
+
"model_type": "siglip_vision_model",
|
| 106 |
+
"num_attention_heads": 12,
|
| 107 |
+
"num_channels": 3,
|
| 108 |
+
"num_hidden_layers": 12,
|
| 109 |
+
"patch_size": 16,
|
| 110 |
+
"torch_dtype": "float32"
|
| 111 |
+
}
|
| 112 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e396dbc84e31c71200f42dd0ba85793307bc33ef3ef0d03c42a8f07724264853
|
| 3 |
+
size 371669508
|
preprocessor_config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"do_convert_rgb": null,
|
| 3 |
+
"do_normalize": true,
|
| 4 |
+
"do_rescale": true,
|
| 5 |
+
"do_resize": true,
|
| 6 |
+
"image_mean": [
|
| 7 |
+
0.5,
|
| 8 |
+
0.5,
|
| 9 |
+
0.5
|
| 10 |
+
],
|
| 11 |
+
"image_processor_type": "SiglipImageProcessor",
|
| 12 |
+
"image_std": [
|
| 13 |
+
0.5,
|
| 14 |
+
0.5,
|
| 15 |
+
0.5
|
| 16 |
+
],
|
| 17 |
+
"processor_class": "SiglipProcessor",
|
| 18 |
+
"resample": 2,
|
| 19 |
+
"rescale_factor": 0.00392156862745098,
|
| 20 |
+
"size": {
|
| 21 |
+
"height": 224,
|
| 22 |
+
"width": 224
|
| 23 |
+
}
|
| 24 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12dd34e788f5eb534c4b6ac8f57ace245fefa2a536daa2fe035c88d0385d96e8
|
| 3 |
+
size 5304
|