Training in progress, epoch 1

config.json

@@ -0,0 +1,188 @@
+{
+  "architectures": [
+    "ASTForAudioClassification"
+  ],
+  "attention_probs_dropout_prob": 0.0,
+  "frequency_stride": 10,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.0,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "_silence_",
+    "1": "_unknown_",
+    "10": "car",
+    "11": "cat",
+    "12": "doctor",
+    "13": "dog",
+    "14": "door",
+    "15": "ear",
+    "16": "egg",
+    "17": "elephant",
+    "18": "fan",
+    "19": "fish",
+    "2": "air",
+    "20": "fox",
+    "21": "frog",
+    "22": "game",
+    "23": "gate",
+    "24": "gold",
+    "25": "hand",
+    "26": "hat",
+    "27": "home",
+    "28": "ice",
+    "29": "ill",
+    "3": "apple",
+    "30": "iron",
+    "31": "jelly",
+    "32": "juice",
+    "33": "jump",
+    "34": "key",
+    "35": "kid",
+    "36": "kite",
+    "37": "lamp",
+    "38": "leaf",
+    "39": "lion",
+    "4": "arm",
+    "40": "milk",
+    "41": "monkey",
+    "42": "moon",
+    "43": "nest",
+    "44": "net",
+    "45": "nose",
+    "46": "ocean",
+    "47": "open",
+    "48": "orange",
+    "49": "pen",
+    "5": "ball",
+    "50": "pig",
+    "51": "pizza",
+    "52": "queen",
+    "53": "question",
+    "54": "quiet",
+    "55": "red",
+    "56": "room",
+    "57": "run",
+    "58": "six",
+    "59": "snake",
+    "6": "bear",
+    "60": "star",
+    "61": "sun",
+    "62": "tiger",
+    "63": "toy",
+    "64": "tree",
+    "65": "umbrella",
+    "66": "under",
+    "67": "unlike",
+    "68": "van",
+    "69": "vest",
+    "7": "bed",
+    "70": "violin",
+    "71": "water",
+    "72": "white",
+    "73": "window",
+    "74": "year",
+    "75": "yellow",
+    "76": "youth",
+    "77": "zero",
+    "78": "zone",
+    "79": "zoo",
+    "8": "box",
+    "9": "cake"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "_silence_": "0",
+    "_unknown_": "1",
+    "air": "2",
+    "apple": "3",
+    "arm": "4",
+    "ball": "5",
+    "bear": "6",
+    "bed": "7",
+    "box": "8",
+    "cake": "9",
+    "car": "10",
+    "cat": "11",
+    "doctor": "12",
+    "dog": "13",
+    "door": "14",
+    "ear": "15",
+    "egg": "16",
+    "elephant": "17",
+    "fan": "18",
+    "fish": "19",
+    "fox": "20",
+    "frog": "21",
+    "game": "22",
+    "gate": "23",
+    "gold": "24",
+    "hand": "25",
+    "hat": "26",
+    "home": "27",
+    "ice": "28",
+    "ill": "29",
+    "iron": "30",
+    "jelly": "31",
+    "juice": "32",
+    "jump": "33",
+    "key": "34",
+    "kid": "35",
+    "kite": "36",
+    "lamp": "37",
+    "leaf": "38",
+    "lion": "39",
+    "milk": "40",
+    "monkey": "41",
+    "moon": "42",
+    "nest": "43",
+    "net": "44",
+    "nose": "45",
+    "ocean": "46",
+    "open": "47",
+    "orange": "48",
+    "pen": "49",
+    "pig": "50",
+    "pizza": "51",
+    "queen": "52",
+    "question": "53",
+    "quiet": "54",
+    "red": "55",
+    "room": "56",
+    "run": "57",
+    "six": "58",
+    "snake": "59",
+    "star": "60",
+    "sun": "61",
+    "tiger": "62",
+    "toy": "63",
+    "tree": "64",
+    "umbrella": "65",
+    "under": "66",
+    "unlike": "67",
+    "van": "68",
+    "vest": "69",
+    "violin": "70",
+    "water": "71",
+    "white": "72",
+    "window": "73",
+    "year": "74",
+    "yellow": "75",
+    "youth": "76",
+    "zero": "77",
+    "zone": "78",
+    "zoo": "79"
+  },
+  "layer_norm_eps": 1e-12,
+  "max_length": 128,
+  "model_type": "audio-spectrogram-transformer",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "num_mel_bins": 128,
+  "patch_size": 16,
+  "problem_type": "single_label_classification",
+  "qkv_bias": true,
+  "time_stride": 10,
+  "torch_dtype": "float32",
+  "transformers_version": "4.51.3"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4aa269d9d57518e1d581862fbc875028afb47404a740b41939ab327e0a3a830
+size 341749072

preprocessor_config.json

@@ -0,0 +1,13 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "ASTFeatureExtractor",
+  "feature_size": 1,
+  "max_length": 128,
+  "mean": -6.845978,
+  "num_mel_bins": 128,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": false,
+  "sampling_rate": 16000,
+  "std": 5.5654526
+}

runs/May15_16-26-02_ip-10-192-11-251/events.out.tfevents.1747326382.ip-10-192-11-251.3223.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:155572595c3dc11d02a34d5ae540f6d0fd64494241d2ab066b80c42c43878079
+size 77200

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a069d67b065ce361b1b3f330f79ee54299b538bba0cd4d23fbbdde581a5e7d95
+size 5841