Upload folder using huggingface_hub
Browse files- checkpoint-1970/config.json +132 -0
- checkpoint-1970/model.safetensors +3 -0
- checkpoint-1970/optimizer.pt +3 -0
- checkpoint-1970/preprocessor_config.json +24 -0
- checkpoint-1970/rng_state.pth +3 -0
- checkpoint-1970/scheduler.pt +3 -0
- checkpoint-1970/trainer_state.json +75 -0
- checkpoint-1970/training_args.bin +3 -0
- checkpoint-985/config.json +132 -0
- checkpoint-985/model.safetensors +3 -0
- checkpoint-985/optimizer.pt +3 -0
- checkpoint-985/preprocessor_config.json +24 -0
- checkpoint-985/rng_state.pth +3 -0
- checkpoint-985/scheduler.pt +3 -0
- checkpoint-985/trainer_state.json +51 -0
- checkpoint-985/training_args.bin +3 -0
- config.json +132 -0
- model.safetensors +3 -0
- preprocessor_config.json +24 -0
- training_args.bin +3 -0
checkpoint-1970/config.json
ADDED
|
@@ -0,0 +1,132 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"SiglipForImageClassification"
|
| 4 |
+
],
|
| 5 |
+
"id2label": {
|
| 6 |
+
"0": "airplane",
|
| 7 |
+
"1": "airport",
|
| 8 |
+
"2": "baseball diamond",
|
| 9 |
+
"3": "basketball court",
|
| 10 |
+
"4": "beach",
|
| 11 |
+
"5": "bridge",
|
| 12 |
+
"6": "chaparral",
|
| 13 |
+
"7": "church",
|
| 14 |
+
"8": "circular farmland",
|
| 15 |
+
"9": "cloud",
|
| 16 |
+
"10": "commercial area",
|
| 17 |
+
"11": "dense residential",
|
| 18 |
+
"12": "desert",
|
| 19 |
+
"13": "forest",
|
| 20 |
+
"14": "freeway",
|
| 21 |
+
"15": "golf course",
|
| 22 |
+
"16": "ground track field",
|
| 23 |
+
"17": "harbor",
|
| 24 |
+
"18": "industrial area",
|
| 25 |
+
"19": "intersection",
|
| 26 |
+
"20": "island",
|
| 27 |
+
"21": "lake",
|
| 28 |
+
"22": "meadow",
|
| 29 |
+
"23": "medium residential",
|
| 30 |
+
"24": "mobile home park",
|
| 31 |
+
"25": "mountain",
|
| 32 |
+
"26": "overpass",
|
| 33 |
+
"27": "palace",
|
| 34 |
+
"28": "parking lot",
|
| 35 |
+
"29": "railway",
|
| 36 |
+
"30": "railway station",
|
| 37 |
+
"31": "rectangular farmland",
|
| 38 |
+
"32": "river",
|
| 39 |
+
"33": "roundabout",
|
| 40 |
+
"34": "runway",
|
| 41 |
+
"35": "sea ice",
|
| 42 |
+
"36": "ship",
|
| 43 |
+
"37": "snowberg",
|
| 44 |
+
"38": "sparse residential",
|
| 45 |
+
"39": "stadium",
|
| 46 |
+
"40": "storage tank",
|
| 47 |
+
"41": "tennis court",
|
| 48 |
+
"42": "terrace",
|
| 49 |
+
"43": "thermal power station",
|
| 50 |
+
"44": "wetland"
|
| 51 |
+
},
|
| 52 |
+
"initializer_factor": 1.0,
|
| 53 |
+
"label2id": {
|
| 54 |
+
"airplane": 0,
|
| 55 |
+
"airport": 1,
|
| 56 |
+
"baseball diamond": 2,
|
| 57 |
+
"basketball court": 3,
|
| 58 |
+
"beach": 4,
|
| 59 |
+
"bridge": 5,
|
| 60 |
+
"chaparral": 6,
|
| 61 |
+
"church": 7,
|
| 62 |
+
"circular farmland": 8,
|
| 63 |
+
"cloud": 9,
|
| 64 |
+
"commercial area": 10,
|
| 65 |
+
"dense residential": 11,
|
| 66 |
+
"desert": 12,
|
| 67 |
+
"forest": 13,
|
| 68 |
+
"freeway": 14,
|
| 69 |
+
"golf course": 15,
|
| 70 |
+
"ground track field": 16,
|
| 71 |
+
"harbor": 17,
|
| 72 |
+
"industrial area": 18,
|
| 73 |
+
"intersection": 19,
|
| 74 |
+
"island": 20,
|
| 75 |
+
"lake": 21,
|
| 76 |
+
"meadow": 22,
|
| 77 |
+
"medium residential": 23,
|
| 78 |
+
"mobile home park": 24,
|
| 79 |
+
"mountain": 25,
|
| 80 |
+
"overpass": 26,
|
| 81 |
+
"palace": 27,
|
| 82 |
+
"parking lot": 28,
|
| 83 |
+
"railway": 29,
|
| 84 |
+
"railway station": 30,
|
| 85 |
+
"rectangular farmland": 31,
|
| 86 |
+
"river": 32,
|
| 87 |
+
"roundabout": 33,
|
| 88 |
+
"runway": 34,
|
| 89 |
+
"sea ice": 35,
|
| 90 |
+
"ship": 36,
|
| 91 |
+
"snowberg": 37,
|
| 92 |
+
"sparse residential": 38,
|
| 93 |
+
"stadium": 39,
|
| 94 |
+
"storage tank": 40,
|
| 95 |
+
"tennis court": 41,
|
| 96 |
+
"terrace": 42,
|
| 97 |
+
"thermal power station": 43,
|
| 98 |
+
"wetland": 44
|
| 99 |
+
},
|
| 100 |
+
"model_type": "siglip",
|
| 101 |
+
"problem_type": "single_label_classification",
|
| 102 |
+
"text_config": {
|
| 103 |
+
"attention_dropout": 0.0,
|
| 104 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 105 |
+
"hidden_size": 768,
|
| 106 |
+
"intermediate_size": 3072,
|
| 107 |
+
"layer_norm_eps": 1e-06,
|
| 108 |
+
"max_position_embeddings": 64,
|
| 109 |
+
"model_type": "siglip_text_model",
|
| 110 |
+
"num_attention_heads": 12,
|
| 111 |
+
"num_hidden_layers": 12,
|
| 112 |
+
"projection_size": 768,
|
| 113 |
+
"torch_dtype": "float32",
|
| 114 |
+
"vocab_size": 256000
|
| 115 |
+
},
|
| 116 |
+
"torch_dtype": "float32",
|
| 117 |
+
"transformers_version": "4.50.0",
|
| 118 |
+
"vision_config": {
|
| 119 |
+
"attention_dropout": 0.0,
|
| 120 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 121 |
+
"hidden_size": 768,
|
| 122 |
+
"image_size": 224,
|
| 123 |
+
"intermediate_size": 3072,
|
| 124 |
+
"layer_norm_eps": 1e-06,
|
| 125 |
+
"model_type": "siglip_vision_model",
|
| 126 |
+
"num_attention_heads": 12,
|
| 127 |
+
"num_channels": 3,
|
| 128 |
+
"num_hidden_layers": 12,
|
| 129 |
+
"patch_size": 16,
|
| 130 |
+
"torch_dtype": "float32"
|
| 131 |
+
}
|
| 132 |
+
}
|
checkpoint-1970/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:de7735b9c19f11a56c6b5f961a56bbf3642ab81f80f42667c04d85b912688995
|
| 3 |
+
size 371700268
|
checkpoint-1970/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:26ae95bbd3ad4380821502da8269b682ac908a13704ad9681fe6fe2e7df95ac4
|
| 3 |
+
size 686820218
|
checkpoint-1970/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-1970/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9b6759887a343b9a59e49e3e78978b8dbec94acefcbc28d0c9a26661bf7c8325
|
| 3 |
+
size 14244
|
checkpoint-1970/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:17410f7b586375e7f0144698da2a32380c4ff4c732cbf740c0889ec436006593
|
| 3 |
+
size 1064
|
checkpoint-1970/trainer_state.json
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 1970,
|
| 3 |
+
"best_metric": 0.13895583152770996,
|
| 4 |
+
"best_model_checkpoint": "siglip2-finetune-full/checkpoint-1970",
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1970,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.5076142131979695,
|
| 14 |
+
"grad_norm": 8.941224098205566,
|
| 15 |
+
"learning_rate": 0.000153125,
|
| 16 |
+
"loss": 1.4212,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 1.0,
|
| 21 |
+
"eval_accuracy": 0.8340634920634921,
|
| 22 |
+
"eval_loss": 0.5306099653244019,
|
| 23 |
+
"eval_model_preparation_time": 0.003,
|
| 24 |
+
"eval_runtime": 415.7745,
|
| 25 |
+
"eval_samples_per_second": 75.762,
|
| 26 |
+
"eval_steps_per_second": 9.471,
|
| 27 |
+
"step": 985
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 1.015228426395939,
|
| 31 |
+
"grad_norm": 8.13437557220459,
|
| 32 |
+
"learning_rate": 0.00010104166666666668,
|
| 33 |
+
"loss": 0.686,
|
| 34 |
+
"step": 1000
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"epoch": 1.5228426395939088,
|
| 38 |
+
"grad_norm": 7.881953239440918,
|
| 39 |
+
"learning_rate": 4.8958333333333335e-05,
|
| 40 |
+
"loss": 0.3992,
|
| 41 |
+
"step": 1500
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 2.0,
|
| 45 |
+
"eval_accuracy": 0.9532063492063492,
|
| 46 |
+
"eval_loss": 0.13895583152770996,
|
| 47 |
+
"eval_model_preparation_time": 0.003,
|
| 48 |
+
"eval_runtime": 418.2101,
|
| 49 |
+
"eval_samples_per_second": 75.321,
|
| 50 |
+
"eval_steps_per_second": 9.416,
|
| 51 |
+
"step": 1970
|
| 52 |
+
}
|
| 53 |
+
],
|
| 54 |
+
"logging_steps": 500,
|
| 55 |
+
"max_steps": 1970,
|
| 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": 5.278477735286784e+18,
|
| 72 |
+
"train_batch_size": 32,
|
| 73 |
+
"trial_name": null,
|
| 74 |
+
"trial_params": null
|
| 75 |
+
}
|
checkpoint-1970/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5371dc4f3b7759a694db9f7dd1b77d5c2e28b4074a617a682427c213e9fb3ec6
|
| 3 |
+
size 5304
|
checkpoint-985/config.json
ADDED
|
@@ -0,0 +1,132 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"SiglipForImageClassification"
|
| 4 |
+
],
|
| 5 |
+
"id2label": {
|
| 6 |
+
"0": "airplane",
|
| 7 |
+
"1": "airport",
|
| 8 |
+
"2": "baseball diamond",
|
| 9 |
+
"3": "basketball court",
|
| 10 |
+
"4": "beach",
|
| 11 |
+
"5": "bridge",
|
| 12 |
+
"6": "chaparral",
|
| 13 |
+
"7": "church",
|
| 14 |
+
"8": "circular farmland",
|
| 15 |
+
"9": "cloud",
|
| 16 |
+
"10": "commercial area",
|
| 17 |
+
"11": "dense residential",
|
| 18 |
+
"12": "desert",
|
| 19 |
+
"13": "forest",
|
| 20 |
+
"14": "freeway",
|
| 21 |
+
"15": "golf course",
|
| 22 |
+
"16": "ground track field",
|
| 23 |
+
"17": "harbor",
|
| 24 |
+
"18": "industrial area",
|
| 25 |
+
"19": "intersection",
|
| 26 |
+
"20": "island",
|
| 27 |
+
"21": "lake",
|
| 28 |
+
"22": "meadow",
|
| 29 |
+
"23": "medium residential",
|
| 30 |
+
"24": "mobile home park",
|
| 31 |
+
"25": "mountain",
|
| 32 |
+
"26": "overpass",
|
| 33 |
+
"27": "palace",
|
| 34 |
+
"28": "parking lot",
|
| 35 |
+
"29": "railway",
|
| 36 |
+
"30": "railway station",
|
| 37 |
+
"31": "rectangular farmland",
|
| 38 |
+
"32": "river",
|
| 39 |
+
"33": "roundabout",
|
| 40 |
+
"34": "runway",
|
| 41 |
+
"35": "sea ice",
|
| 42 |
+
"36": "ship",
|
| 43 |
+
"37": "snowberg",
|
| 44 |
+
"38": "sparse residential",
|
| 45 |
+
"39": "stadium",
|
| 46 |
+
"40": "storage tank",
|
| 47 |
+
"41": "tennis court",
|
| 48 |
+
"42": "terrace",
|
| 49 |
+
"43": "thermal power station",
|
| 50 |
+
"44": "wetland"
|
| 51 |
+
},
|
| 52 |
+
"initializer_factor": 1.0,
|
| 53 |
+
"label2id": {
|
| 54 |
+
"airplane": 0,
|
| 55 |
+
"airport": 1,
|
| 56 |
+
"baseball diamond": 2,
|
| 57 |
+
"basketball court": 3,
|
| 58 |
+
"beach": 4,
|
| 59 |
+
"bridge": 5,
|
| 60 |
+
"chaparral": 6,
|
| 61 |
+
"church": 7,
|
| 62 |
+
"circular farmland": 8,
|
| 63 |
+
"cloud": 9,
|
| 64 |
+
"commercial area": 10,
|
| 65 |
+
"dense residential": 11,
|
| 66 |
+
"desert": 12,
|
| 67 |
+
"forest": 13,
|
| 68 |
+
"freeway": 14,
|
| 69 |
+
"golf course": 15,
|
| 70 |
+
"ground track field": 16,
|
| 71 |
+
"harbor": 17,
|
| 72 |
+
"industrial area": 18,
|
| 73 |
+
"intersection": 19,
|
| 74 |
+
"island": 20,
|
| 75 |
+
"lake": 21,
|
| 76 |
+
"meadow": 22,
|
| 77 |
+
"medium residential": 23,
|
| 78 |
+
"mobile home park": 24,
|
| 79 |
+
"mountain": 25,
|
| 80 |
+
"overpass": 26,
|
| 81 |
+
"palace": 27,
|
| 82 |
+
"parking lot": 28,
|
| 83 |
+
"railway": 29,
|
| 84 |
+
"railway station": 30,
|
| 85 |
+
"rectangular farmland": 31,
|
| 86 |
+
"river": 32,
|
| 87 |
+
"roundabout": 33,
|
| 88 |
+
"runway": 34,
|
| 89 |
+
"sea ice": 35,
|
| 90 |
+
"ship": 36,
|
| 91 |
+
"snowberg": 37,
|
| 92 |
+
"sparse residential": 38,
|
| 93 |
+
"stadium": 39,
|
| 94 |
+
"storage tank": 40,
|
| 95 |
+
"tennis court": 41,
|
| 96 |
+
"terrace": 42,
|
| 97 |
+
"thermal power station": 43,
|
| 98 |
+
"wetland": 44
|
| 99 |
+
},
|
| 100 |
+
"model_type": "siglip",
|
| 101 |
+
"problem_type": "single_label_classification",
|
| 102 |
+
"text_config": {
|
| 103 |
+
"attention_dropout": 0.0,
|
| 104 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 105 |
+
"hidden_size": 768,
|
| 106 |
+
"intermediate_size": 3072,
|
| 107 |
+
"layer_norm_eps": 1e-06,
|
| 108 |
+
"max_position_embeddings": 64,
|
| 109 |
+
"model_type": "siglip_text_model",
|
| 110 |
+
"num_attention_heads": 12,
|
| 111 |
+
"num_hidden_layers": 12,
|
| 112 |
+
"projection_size": 768,
|
| 113 |
+
"torch_dtype": "float32",
|
| 114 |
+
"vocab_size": 256000
|
| 115 |
+
},
|
| 116 |
+
"torch_dtype": "float32",
|
| 117 |
+
"transformers_version": "4.50.0",
|
| 118 |
+
"vision_config": {
|
| 119 |
+
"attention_dropout": 0.0,
|
| 120 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 121 |
+
"hidden_size": 768,
|
| 122 |
+
"image_size": 224,
|
| 123 |
+
"intermediate_size": 3072,
|
| 124 |
+
"layer_norm_eps": 1e-06,
|
| 125 |
+
"model_type": "siglip_vision_model",
|
| 126 |
+
"num_attention_heads": 12,
|
| 127 |
+
"num_channels": 3,
|
| 128 |
+
"num_hidden_layers": 12,
|
| 129 |
+
"patch_size": 16,
|
| 130 |
+
"torch_dtype": "float32"
|
| 131 |
+
}
|
| 132 |
+
}
|
checkpoint-985/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:55af477d364fa49ebf55166850ad91444a8211c41d17f111249b0ad40c588b0f
|
| 3 |
+
size 371700268
|
checkpoint-985/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1355263d8f83f260a7aa62e7c4130be86af2a237592e67960054e3e7e2ad031
|
| 3 |
+
size 686820218
|
checkpoint-985/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-985/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d03065d21ca2e32ab40ef969161b89aacc21f83bf3df69a74f0df011f4954660
|
| 3 |
+
size 14244
|
checkpoint-985/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:248877ee4d1e62307eb7d45650f10cdca8991a166a68f8918d0986a6c12d748e
|
| 3 |
+
size 1064
|
checkpoint-985/trainer_state.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 985,
|
| 3 |
+
"best_metric": 0.5306099653244019,
|
| 4 |
+
"best_model_checkpoint": "siglip2-finetune-full/checkpoint-985",
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 985,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.5076142131979695,
|
| 14 |
+
"grad_norm": 8.941224098205566,
|
| 15 |
+
"learning_rate": 0.000153125,
|
| 16 |
+
"loss": 1.4212,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 1.0,
|
| 21 |
+
"eval_accuracy": 0.8340634920634921,
|
| 22 |
+
"eval_loss": 0.5306099653244019,
|
| 23 |
+
"eval_model_preparation_time": 0.003,
|
| 24 |
+
"eval_runtime": 415.7745,
|
| 25 |
+
"eval_samples_per_second": 75.762,
|
| 26 |
+
"eval_steps_per_second": 9.471,
|
| 27 |
+
"step": 985
|
| 28 |
+
}
|
| 29 |
+
],
|
| 30 |
+
"logging_steps": 500,
|
| 31 |
+
"max_steps": 1970,
|
| 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": 2.639238867643392e+18,
|
| 48 |
+
"train_batch_size": 32,
|
| 49 |
+
"trial_name": null,
|
| 50 |
+
"trial_params": null
|
| 51 |
+
}
|
checkpoint-985/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5371dc4f3b7759a694db9f7dd1b77d5c2e28b4074a617a682427c213e9fb3ec6
|
| 3 |
+
size 5304
|
config.json
ADDED
|
@@ -0,0 +1,132 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"SiglipForImageClassification"
|
| 4 |
+
],
|
| 5 |
+
"id2label": {
|
| 6 |
+
"0": "airplane",
|
| 7 |
+
"1": "airport",
|
| 8 |
+
"2": "baseball diamond",
|
| 9 |
+
"3": "basketball court",
|
| 10 |
+
"4": "beach",
|
| 11 |
+
"5": "bridge",
|
| 12 |
+
"6": "chaparral",
|
| 13 |
+
"7": "church",
|
| 14 |
+
"8": "circular farmland",
|
| 15 |
+
"9": "cloud",
|
| 16 |
+
"10": "commercial area",
|
| 17 |
+
"11": "dense residential",
|
| 18 |
+
"12": "desert",
|
| 19 |
+
"13": "forest",
|
| 20 |
+
"14": "freeway",
|
| 21 |
+
"15": "golf course",
|
| 22 |
+
"16": "ground track field",
|
| 23 |
+
"17": "harbor",
|
| 24 |
+
"18": "industrial area",
|
| 25 |
+
"19": "intersection",
|
| 26 |
+
"20": "island",
|
| 27 |
+
"21": "lake",
|
| 28 |
+
"22": "meadow",
|
| 29 |
+
"23": "medium residential",
|
| 30 |
+
"24": "mobile home park",
|
| 31 |
+
"25": "mountain",
|
| 32 |
+
"26": "overpass",
|
| 33 |
+
"27": "palace",
|
| 34 |
+
"28": "parking lot",
|
| 35 |
+
"29": "railway",
|
| 36 |
+
"30": "railway station",
|
| 37 |
+
"31": "rectangular farmland",
|
| 38 |
+
"32": "river",
|
| 39 |
+
"33": "roundabout",
|
| 40 |
+
"34": "runway",
|
| 41 |
+
"35": "sea ice",
|
| 42 |
+
"36": "ship",
|
| 43 |
+
"37": "snowberg",
|
| 44 |
+
"38": "sparse residential",
|
| 45 |
+
"39": "stadium",
|
| 46 |
+
"40": "storage tank",
|
| 47 |
+
"41": "tennis court",
|
| 48 |
+
"42": "terrace",
|
| 49 |
+
"43": "thermal power station",
|
| 50 |
+
"44": "wetland"
|
| 51 |
+
},
|
| 52 |
+
"initializer_factor": 1.0,
|
| 53 |
+
"label2id": {
|
| 54 |
+
"airplane": 0,
|
| 55 |
+
"airport": 1,
|
| 56 |
+
"baseball diamond": 2,
|
| 57 |
+
"basketball court": 3,
|
| 58 |
+
"beach": 4,
|
| 59 |
+
"bridge": 5,
|
| 60 |
+
"chaparral": 6,
|
| 61 |
+
"church": 7,
|
| 62 |
+
"circular farmland": 8,
|
| 63 |
+
"cloud": 9,
|
| 64 |
+
"commercial area": 10,
|
| 65 |
+
"dense residential": 11,
|
| 66 |
+
"desert": 12,
|
| 67 |
+
"forest": 13,
|
| 68 |
+
"freeway": 14,
|
| 69 |
+
"golf course": 15,
|
| 70 |
+
"ground track field": 16,
|
| 71 |
+
"harbor": 17,
|
| 72 |
+
"industrial area": 18,
|
| 73 |
+
"intersection": 19,
|
| 74 |
+
"island": 20,
|
| 75 |
+
"lake": 21,
|
| 76 |
+
"meadow": 22,
|
| 77 |
+
"medium residential": 23,
|
| 78 |
+
"mobile home park": 24,
|
| 79 |
+
"mountain": 25,
|
| 80 |
+
"overpass": 26,
|
| 81 |
+
"palace": 27,
|
| 82 |
+
"parking lot": 28,
|
| 83 |
+
"railway": 29,
|
| 84 |
+
"railway station": 30,
|
| 85 |
+
"rectangular farmland": 31,
|
| 86 |
+
"river": 32,
|
| 87 |
+
"roundabout": 33,
|
| 88 |
+
"runway": 34,
|
| 89 |
+
"sea ice": 35,
|
| 90 |
+
"ship": 36,
|
| 91 |
+
"snowberg": 37,
|
| 92 |
+
"sparse residential": 38,
|
| 93 |
+
"stadium": 39,
|
| 94 |
+
"storage tank": 40,
|
| 95 |
+
"tennis court": 41,
|
| 96 |
+
"terrace": 42,
|
| 97 |
+
"thermal power station": 43,
|
| 98 |
+
"wetland": 44
|
| 99 |
+
},
|
| 100 |
+
"model_type": "siglip",
|
| 101 |
+
"problem_type": "single_label_classification",
|
| 102 |
+
"text_config": {
|
| 103 |
+
"attention_dropout": 0.0,
|
| 104 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 105 |
+
"hidden_size": 768,
|
| 106 |
+
"intermediate_size": 3072,
|
| 107 |
+
"layer_norm_eps": 1e-06,
|
| 108 |
+
"max_position_embeddings": 64,
|
| 109 |
+
"model_type": "siglip_text_model",
|
| 110 |
+
"num_attention_heads": 12,
|
| 111 |
+
"num_hidden_layers": 12,
|
| 112 |
+
"projection_size": 768,
|
| 113 |
+
"torch_dtype": "float32",
|
| 114 |
+
"vocab_size": 256000
|
| 115 |
+
},
|
| 116 |
+
"torch_dtype": "float32",
|
| 117 |
+
"transformers_version": "4.50.0",
|
| 118 |
+
"vision_config": {
|
| 119 |
+
"attention_dropout": 0.0,
|
| 120 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 121 |
+
"hidden_size": 768,
|
| 122 |
+
"image_size": 224,
|
| 123 |
+
"intermediate_size": 3072,
|
| 124 |
+
"layer_norm_eps": 1e-06,
|
| 125 |
+
"model_type": "siglip_vision_model",
|
| 126 |
+
"num_attention_heads": 12,
|
| 127 |
+
"num_channels": 3,
|
| 128 |
+
"num_hidden_layers": 12,
|
| 129 |
+
"patch_size": 16,
|
| 130 |
+
"torch_dtype": "float32"
|
| 131 |
+
}
|
| 132 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:de7735b9c19f11a56c6b5f961a56bbf3642ab81f80f42667c04d85b912688995
|
| 3 |
+
size 371700268
|
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 sha256:5371dc4f3b7759a694db9f7dd1b77d5c2e28b4074a617a682427c213e9fb3ec6
|
| 3 |
+
size 5304
|