Upload 7 files
Browse files- config.json +120 -0
- model.safetensors +3 -0
- rng_state.pth +3 -0
- scaler.pt +3 -0
- scheduler.pt +3 -0
- trainer_state.json +2610 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,120 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"SegformerForSemanticSegmentation"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.0,
|
| 6 |
+
"classifier_dropout_prob": 0.1,
|
| 7 |
+
"decoder_hidden_size": 768,
|
| 8 |
+
"depths": [
|
| 9 |
+
3,
|
| 10 |
+
4,
|
| 11 |
+
6,
|
| 12 |
+
3
|
| 13 |
+
],
|
| 14 |
+
"downsampling_rates": [
|
| 15 |
+
1,
|
| 16 |
+
4,
|
| 17 |
+
8,
|
| 18 |
+
16
|
| 19 |
+
],
|
| 20 |
+
"drop_path_rate": 0.1,
|
| 21 |
+
"dtype": "float32",
|
| 22 |
+
"hidden_act": "gelu",
|
| 23 |
+
"hidden_dropout_prob": 0.0,
|
| 24 |
+
"hidden_sizes": [
|
| 25 |
+
64,
|
| 26 |
+
128,
|
| 27 |
+
320,
|
| 28 |
+
512
|
| 29 |
+
],
|
| 30 |
+
"id2label": {
|
| 31 |
+
"0": "background",
|
| 32 |
+
"1": "shirt, blouse",
|
| 33 |
+
"2": "top, t-shirt, sweatshirt",
|
| 34 |
+
"3": "sweater",
|
| 35 |
+
"4": "cardigan",
|
| 36 |
+
"5": "jacket",
|
| 37 |
+
"6": "vest",
|
| 38 |
+
"7": "pants",
|
| 39 |
+
"8": "shorts",
|
| 40 |
+
"9": "skirt",
|
| 41 |
+
"10": "coat",
|
| 42 |
+
"11": "dress",
|
| 43 |
+
"12": "glasses",
|
| 44 |
+
"13": "hat",
|
| 45 |
+
"14": "tie",
|
| 46 |
+
"15": "glove",
|
| 47 |
+
"16": "watch",
|
| 48 |
+
"17": "belt",
|
| 49 |
+
"18": "tights, stockings",
|
| 50 |
+
"19": "sock",
|
| 51 |
+
"20": "shoe",
|
| 52 |
+
"21": "bag",
|
| 53 |
+
"22": "scarf"
|
| 54 |
+
},
|
| 55 |
+
"image_size": 224,
|
| 56 |
+
"initializer_range": 0.02,
|
| 57 |
+
"label2id": {
|
| 58 |
+
"background": 0,
|
| 59 |
+
"bag": 21,
|
| 60 |
+
"belt": 17,
|
| 61 |
+
"cardigan": 4,
|
| 62 |
+
"coat": 10,
|
| 63 |
+
"dress": 11,
|
| 64 |
+
"glasses": 12,
|
| 65 |
+
"glove": 15,
|
| 66 |
+
"hat": 13,
|
| 67 |
+
"jacket": 5,
|
| 68 |
+
"pants": 7,
|
| 69 |
+
"scarf": 22,
|
| 70 |
+
"shirt, blouse": 1,
|
| 71 |
+
"shoe": 20,
|
| 72 |
+
"shorts": 8,
|
| 73 |
+
"skirt": 9,
|
| 74 |
+
"sock": 19,
|
| 75 |
+
"sweater": 3,
|
| 76 |
+
"tie": 14,
|
| 77 |
+
"tights, stockings": 18,
|
| 78 |
+
"top, t-shirt, sweatshirt": 2,
|
| 79 |
+
"vest": 6,
|
| 80 |
+
"watch": 16
|
| 81 |
+
},
|
| 82 |
+
"layer_norm_eps": 1e-06,
|
| 83 |
+
"mlp_ratios": [
|
| 84 |
+
4,
|
| 85 |
+
4,
|
| 86 |
+
4,
|
| 87 |
+
4
|
| 88 |
+
],
|
| 89 |
+
"model_type": "segformer",
|
| 90 |
+
"num_attention_heads": [
|
| 91 |
+
1,
|
| 92 |
+
2,
|
| 93 |
+
5,
|
| 94 |
+
8
|
| 95 |
+
],
|
| 96 |
+
"num_channels": 3,
|
| 97 |
+
"num_encoder_blocks": 4,
|
| 98 |
+
"patch_sizes": [
|
| 99 |
+
7,
|
| 100 |
+
3,
|
| 101 |
+
3,
|
| 102 |
+
3
|
| 103 |
+
],
|
| 104 |
+
"reshape_last_stage": true,
|
| 105 |
+
"semantic_loss_ignore_index": 255,
|
| 106 |
+
"sr_ratios": [
|
| 107 |
+
8,
|
| 108 |
+
4,
|
| 109 |
+
2,
|
| 110 |
+
1
|
| 111 |
+
],
|
| 112 |
+
"strides": [
|
| 113 |
+
4,
|
| 114 |
+
2,
|
| 115 |
+
2,
|
| 116 |
+
2
|
| 117 |
+
],
|
| 118 |
+
"transformers_version": "5.0.0",
|
| 119 |
+
"use_cache": false
|
| 120 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d621658b537511c01f033bfa911027fd5996acd6704c1cf8678aa466d7e9b076
|
| 3 |
+
size 109508620
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cbdd6a8933cbfd69bf19d929214ee151a1c09d43a4a2bae6f253cffcd638b67c
|
| 3 |
+
size 14645
|
scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:22632e75595ab0de15567011a5b81255ea87b6ab8abcb24422372dfe2b463975
|
| 3 |
+
size 1383
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0d03d8b37f0a8d6909b2615f9dd1626e10dbd2bf960298ae3245c58abe575d55
|
| 3 |
+
size 1465
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2610 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 4.930486462689853,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 28000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.017609509134932862,
|
| 14 |
+
"grad_norm": 48.91790771484375,
|
| 15 |
+
"learning_rate": 5.989540412044374e-05,
|
| 16 |
+
"loss": 9.243912963867187,
|
| 17 |
+
"step": 100
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.035219018269865725,
|
| 21 |
+
"grad_norm": 10.053231239318848,
|
| 22 |
+
"learning_rate": 5.978975171685156e-05,
|
| 23 |
+
"loss": 3.7935104370117188,
|
| 24 |
+
"step": 200
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.05282852740479859,
|
| 28 |
+
"grad_norm": 10.990548133850098,
|
| 29 |
+
"learning_rate": 5.968409931325938e-05,
|
| 30 |
+
"loss": 3.084054260253906,
|
| 31 |
+
"step": 300
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.07043803653973145,
|
| 35 |
+
"grad_norm": 25.284229278564453,
|
| 36 |
+
"learning_rate": 5.9578446909667196e-05,
|
| 37 |
+
"loss": 2.5839889526367186,
|
| 38 |
+
"step": 400
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08804754567466432,
|
| 42 |
+
"grad_norm": 19.381221771240234,
|
| 43 |
+
"learning_rate": 5.9472794506075014e-05,
|
| 44 |
+
"loss": 2.5452081298828126,
|
| 45 |
+
"step": 500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.08804754567466432,
|
| 49 |
+
"eval_loss": 0.46036240458488464,
|
| 50 |
+
"eval_mIoU": 24.719834678855246,
|
| 51 |
+
"eval_mean_accuracy": 30.302184775266984,
|
| 52 |
+
"eval_overall_accuracy": 93.80432756432208,
|
| 53 |
+
"eval_runtime": 595.2093,
|
| 54 |
+
"eval_samples_per_second": 1.94,
|
| 55 |
+
"eval_steps_per_second": 0.971,
|
| 56 |
+
"step": 500
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.10565705480959718,
|
| 60 |
+
"grad_norm": 30.767061233520508,
|
| 61 |
+
"learning_rate": 5.936714210248283e-05,
|
| 62 |
+
"loss": 2.543914794921875,
|
| 63 |
+
"step": 600
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.12326656394453005,
|
| 67 |
+
"grad_norm": 12.711847305297852,
|
| 68 |
+
"learning_rate": 5.926148969889066e-05,
|
| 69 |
+
"loss": 2.555495300292969,
|
| 70 |
+
"step": 700
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.1408760730794629,
|
| 74 |
+
"grad_norm": 9.562529563903809,
|
| 75 |
+
"learning_rate": 5.915583729529847e-05,
|
| 76 |
+
"loss": 2.357043914794922,
|
| 77 |
+
"step": 800
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 0.15848558221439576,
|
| 81 |
+
"grad_norm": 7.863616943359375,
|
| 82 |
+
"learning_rate": 5.905018489170629e-05,
|
| 83 |
+
"loss": 2.3811045837402345,
|
| 84 |
+
"step": 900
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.17609509134932863,
|
| 88 |
+
"grad_norm": 19.924156188964844,
|
| 89 |
+
"learning_rate": 5.8944532488114105e-05,
|
| 90 |
+
"loss": 2.2813540649414064,
|
| 91 |
+
"step": 1000
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 0.17609509134932863,
|
| 95 |
+
"eval_loss": 0.4134902060031891,
|
| 96 |
+
"eval_mIoU": 28.85459800222985,
|
| 97 |
+
"eval_mean_accuracy": 36.7015683094026,
|
| 98 |
+
"eval_overall_accuracy": 94.40936266085802,
|
| 99 |
+
"eval_runtime": 581.5599,
|
| 100 |
+
"eval_samples_per_second": 1.986,
|
| 101 |
+
"eval_steps_per_second": 0.994,
|
| 102 |
+
"step": 1000
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.1937046004842615,
|
| 106 |
+
"grad_norm": 9.245338439941406,
|
| 107 |
+
"learning_rate": 5.883888008452192e-05,
|
| 108 |
+
"loss": 2.224588165283203,
|
| 109 |
+
"step": 1100
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.21131410961919436,
|
| 113 |
+
"grad_norm": 28.38551902770996,
|
| 114 |
+
"learning_rate": 5.873322768092974e-05,
|
| 115 |
+
"loss": 2.2024305725097655,
|
| 116 |
+
"step": 1200
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.22892361875412723,
|
| 120 |
+
"grad_norm": 21.654741287231445,
|
| 121 |
+
"learning_rate": 5.862757527733756e-05,
|
| 122 |
+
"loss": 2.333185577392578,
|
| 123 |
+
"step": 1300
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.2465331278890601,
|
| 127 |
+
"grad_norm": 22.906660079956055,
|
| 128 |
+
"learning_rate": 5.8521922873745385e-05,
|
| 129 |
+
"loss": 2.1109620666503908,
|
| 130 |
+
"step": 1400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.26414263702399293,
|
| 134 |
+
"grad_norm": 19.337547302246094,
|
| 135 |
+
"learning_rate": 5.8416270470153196e-05,
|
| 136 |
+
"loss": 2.1600161743164064,
|
| 137 |
+
"step": 1500
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.26414263702399293,
|
| 141 |
+
"eval_loss": 0.41332295536994934,
|
| 142 |
+
"eval_mIoU": 30.867547066481954,
|
| 143 |
+
"eval_mean_accuracy": 38.46061833204942,
|
| 144 |
+
"eval_overall_accuracy": 94.4356427873884,
|
| 145 |
+
"eval_runtime": 582.6409,
|
| 146 |
+
"eval_samples_per_second": 1.982,
|
| 147 |
+
"eval_steps_per_second": 0.992,
|
| 148 |
+
"step": 1500
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 0.2817521461589258,
|
| 152 |
+
"grad_norm": 16.490278244018555,
|
| 153 |
+
"learning_rate": 5.831061806656102e-05,
|
| 154 |
+
"loss": 2.0159757995605467,
|
| 155 |
+
"step": 1600
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.29936165529385866,
|
| 159 |
+
"grad_norm": 17.120445251464844,
|
| 160 |
+
"learning_rate": 5.820496566296883e-05,
|
| 161 |
+
"loss": 2.0717059326171876,
|
| 162 |
+
"step": 1700
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 0.31697116442879153,
|
| 166 |
+
"grad_norm": 12.258465766906738,
|
| 167 |
+
"learning_rate": 5.809931325937665e-05,
|
| 168 |
+
"loss": 2.131772918701172,
|
| 169 |
+
"step": 1800
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 0.3345806735637244,
|
| 173 |
+
"grad_norm": 13.74216079711914,
|
| 174 |
+
"learning_rate": 5.799366085578447e-05,
|
| 175 |
+
"loss": 1.8676652526855468,
|
| 176 |
+
"step": 1900
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.35219018269865726,
|
| 180 |
+
"grad_norm": 25.953657150268555,
|
| 181 |
+
"learning_rate": 5.788800845219229e-05,
|
| 182 |
+
"loss": 1.9515042114257812,
|
| 183 |
+
"step": 2000
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 0.35219018269865726,
|
| 187 |
+
"eval_loss": 0.4337688982486725,
|
| 188 |
+
"eval_mIoU": 32.19576589621381,
|
| 189 |
+
"eval_mean_accuracy": 40.59080396049274,
|
| 190 |
+
"eval_overall_accuracy": 94.54845279842228,
|
| 191 |
+
"eval_runtime": 584.2063,
|
| 192 |
+
"eval_samples_per_second": 1.977,
|
| 193 |
+
"eval_steps_per_second": 0.989,
|
| 194 |
+
"step": 2000
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.3697996918335901,
|
| 198 |
+
"grad_norm": 19.75416374206543,
|
| 199 |
+
"learning_rate": 5.778235604860011e-05,
|
| 200 |
+
"loss": 1.784864959716797,
|
| 201 |
+
"step": 2100
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.387409200968523,
|
| 205 |
+
"grad_norm": 22.587085723876953,
|
| 206 |
+
"learning_rate": 5.767670364500792e-05,
|
| 207 |
+
"loss": 1.9940145874023438,
|
| 208 |
+
"step": 2200
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.40501871010345586,
|
| 212 |
+
"grad_norm": 35.613121032714844,
|
| 213 |
+
"learning_rate": 5.757105124141575e-05,
|
| 214 |
+
"loss": 1.9241729736328126,
|
| 215 |
+
"step": 2300
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.4226282192383887,
|
| 219 |
+
"grad_norm": 38.178104400634766,
|
| 220 |
+
"learning_rate": 5.746539883782356e-05,
|
| 221 |
+
"loss": 1.7019007873535157,
|
| 222 |
+
"step": 2400
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.4402377283733216,
|
| 226 |
+
"grad_norm": 35.151058197021484,
|
| 227 |
+
"learning_rate": 5.735974643423138e-05,
|
| 228 |
+
"loss": 1.8641450500488281,
|
| 229 |
+
"step": 2500
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 0.4402377283733216,
|
| 233 |
+
"eval_loss": 0.39942070841789246,
|
| 234 |
+
"eval_mIoU": 33.685957807527856,
|
| 235 |
+
"eval_mean_accuracy": 44.48807007726972,
|
| 236 |
+
"eval_overall_accuracy": 94.60921745795709,
|
| 237 |
+
"eval_runtime": 588.2746,
|
| 238 |
+
"eval_samples_per_second": 1.963,
|
| 239 |
+
"eval_steps_per_second": 0.983,
|
| 240 |
+
"step": 2500
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.45784723750825446,
|
| 244 |
+
"grad_norm": 13.256592750549316,
|
| 245 |
+
"learning_rate": 5.7254094030639196e-05,
|
| 246 |
+
"loss": 1.7548785400390625,
|
| 247 |
+
"step": 2600
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.4754567466431873,
|
| 251 |
+
"grad_norm": 12.395865440368652,
|
| 252 |
+
"learning_rate": 5.7148441627047014e-05,
|
| 253 |
+
"loss": 1.7571292114257813,
|
| 254 |
+
"step": 2700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.4930662557781202,
|
| 258 |
+
"grad_norm": 10.11026668548584,
|
| 259 |
+
"learning_rate": 5.704278922345484e-05,
|
| 260 |
+
"loss": 1.8158778381347656,
|
| 261 |
+
"step": 2800
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.510675764913053,
|
| 265 |
+
"grad_norm": 37.68540954589844,
|
| 266 |
+
"learning_rate": 5.693713681986265e-05,
|
| 267 |
+
"loss": 1.95743408203125,
|
| 268 |
+
"step": 2900
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.5282852740479859,
|
| 272 |
+
"grad_norm": 6.611830234527588,
|
| 273 |
+
"learning_rate": 5.6831484416270475e-05,
|
| 274 |
+
"loss": 1.9499478149414062,
|
| 275 |
+
"step": 3000
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.5282852740479859,
|
| 279 |
+
"eval_loss": 0.3753623068332672,
|
| 280 |
+
"eval_mIoU": 33.48202504810993,
|
| 281 |
+
"eval_mean_accuracy": 42.84434388722273,
|
| 282 |
+
"eval_overall_accuracy": 94.69698059610474,
|
| 283 |
+
"eval_runtime": 592.7509,
|
| 284 |
+
"eval_samples_per_second": 1.949,
|
| 285 |
+
"eval_steps_per_second": 0.975,
|
| 286 |
+
"step": 3000
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 0.5458947831829187,
|
| 290 |
+
"grad_norm": 11.240772247314453,
|
| 291 |
+
"learning_rate": 5.6725832012678287e-05,
|
| 292 |
+
"loss": 1.9529837036132813,
|
| 293 |
+
"step": 3100
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 0.5635042923178516,
|
| 297 |
+
"grad_norm": 5.816749095916748,
|
| 298 |
+
"learning_rate": 5.662017960908611e-05,
|
| 299 |
+
"loss": 1.7379286193847656,
|
| 300 |
+
"step": 3200
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 0.5811138014527845,
|
| 304 |
+
"grad_norm": 14.43251895904541,
|
| 305 |
+
"learning_rate": 5.651452720549393e-05,
|
| 306 |
+
"loss": 1.878860626220703,
|
| 307 |
+
"step": 3300
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 0.5987233105877173,
|
| 311 |
+
"grad_norm": 17.320858001708984,
|
| 312 |
+
"learning_rate": 5.640887480190174e-05,
|
| 313 |
+
"loss": 1.7022770690917968,
|
| 314 |
+
"step": 3400
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.6163328197226502,
|
| 318 |
+
"grad_norm": 8.217686653137207,
|
| 319 |
+
"learning_rate": 5.6303222398309566e-05,
|
| 320 |
+
"loss": 1.7487423706054688,
|
| 321 |
+
"step": 3500
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.6163328197226502,
|
| 325 |
+
"eval_loss": 0.3824031949043274,
|
| 326 |
+
"eval_mIoU": 35.34702178902146,
|
| 327 |
+
"eval_mean_accuracy": 43.66437157324913,
|
| 328 |
+
"eval_overall_accuracy": 95.11653123995958,
|
| 329 |
+
"eval_runtime": 594.2993,
|
| 330 |
+
"eval_samples_per_second": 1.943,
|
| 331 |
+
"eval_steps_per_second": 0.973,
|
| 332 |
+
"step": 3500
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.6339423288575831,
|
| 336 |
+
"grad_norm": 8.632674217224121,
|
| 337 |
+
"learning_rate": 5.619756999471738e-05,
|
| 338 |
+
"loss": 1.7975865173339844,
|
| 339 |
+
"step": 3600
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.6515518379925159,
|
| 343 |
+
"grad_norm": 12.074219703674316,
|
| 344 |
+
"learning_rate": 5.60919175911252e-05,
|
| 345 |
+
"loss": 1.7007231140136718,
|
| 346 |
+
"step": 3700
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.6691613471274488,
|
| 350 |
+
"grad_norm": 28.738109588623047,
|
| 351 |
+
"learning_rate": 5.5986265187533014e-05,
|
| 352 |
+
"loss": 1.7046678161621094,
|
| 353 |
+
"step": 3800
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.6867708562623817,
|
| 357 |
+
"grad_norm": 17.928030014038086,
|
| 358 |
+
"learning_rate": 5.588061278394084e-05,
|
| 359 |
+
"loss": 1.876953125,
|
| 360 |
+
"step": 3900
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.7043803653973145,
|
| 364 |
+
"grad_norm": 8.591980934143066,
|
| 365 |
+
"learning_rate": 5.577496038034866e-05,
|
| 366 |
+
"loss": 1.8151817321777344,
|
| 367 |
+
"step": 4000
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.7043803653973145,
|
| 371 |
+
"eval_loss": 0.385084331035614,
|
| 372 |
+
"eval_mIoU": 35.32089535040148,
|
| 373 |
+
"eval_mean_accuracy": 44.71839496509271,
|
| 374 |
+
"eval_overall_accuracy": 94.84546578807749,
|
| 375 |
+
"eval_runtime": 594.7804,
|
| 376 |
+
"eval_samples_per_second": 1.942,
|
| 377 |
+
"eval_steps_per_second": 0.972,
|
| 378 |
+
"step": 4000
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.7219898745322474,
|
| 382 |
+
"grad_norm": 28.210174560546875,
|
| 383 |
+
"learning_rate": 5.566930797675647e-05,
|
| 384 |
+
"loss": 1.6734780883789062,
|
| 385 |
+
"step": 4100
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.7395993836671803,
|
| 389 |
+
"grad_norm": 48.83910369873047,
|
| 390 |
+
"learning_rate": 5.556365557316429e-05,
|
| 391 |
+
"loss": 1.6707070922851563,
|
| 392 |
+
"step": 4200
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.7572088928021131,
|
| 396 |
+
"grad_norm": 40.12311935424805,
|
| 397 |
+
"learning_rate": 5.5458003169572105e-05,
|
| 398 |
+
"loss": 1.7087435913085938,
|
| 399 |
+
"step": 4300
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.774818401937046,
|
| 403 |
+
"grad_norm": 6.389533042907715,
|
| 404 |
+
"learning_rate": 5.535235076597993e-05,
|
| 405 |
+
"loss": 1.7182331848144532,
|
| 406 |
+
"step": 4400
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.7924279110719789,
|
| 410 |
+
"grad_norm": 9.220948219299316,
|
| 411 |
+
"learning_rate": 5.524669836238774e-05,
|
| 412 |
+
"loss": 1.7504454040527344,
|
| 413 |
+
"step": 4500
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.7924279110719789,
|
| 417 |
+
"eval_loss": 0.3685193955898285,
|
| 418 |
+
"eval_mIoU": 35.59346252124521,
|
| 419 |
+
"eval_mean_accuracy": 46.61321173585393,
|
| 420 |
+
"eval_overall_accuracy": 95.03028737518179,
|
| 421 |
+
"eval_runtime": 596.9732,
|
| 422 |
+
"eval_samples_per_second": 1.935,
|
| 423 |
+
"eval_steps_per_second": 0.968,
|
| 424 |
+
"step": 4500
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.8100374202069117,
|
| 428 |
+
"grad_norm": 10.630125045776367,
|
| 429 |
+
"learning_rate": 5.5141045958795566e-05,
|
| 430 |
+
"loss": 1.6900584411621093,
|
| 431 |
+
"step": 4600
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.8276469293418446,
|
| 435 |
+
"grad_norm": 21.725378036499023,
|
| 436 |
+
"learning_rate": 5.5035393555203384e-05,
|
| 437 |
+
"loss": 1.5272064208984375,
|
| 438 |
+
"step": 4700
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.8452564384767774,
|
| 442 |
+
"grad_norm": 34.975982666015625,
|
| 443 |
+
"learning_rate": 5.49297411516112e-05,
|
| 444 |
+
"loss": 1.5466600036621094,
|
| 445 |
+
"step": 4800
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.8628659476117103,
|
| 449 |
+
"grad_norm": 10.243609428405762,
|
| 450 |
+
"learning_rate": 5.482408874801902e-05,
|
| 451 |
+
"loss": 1.7738760375976563,
|
| 452 |
+
"step": 4900
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.8804754567466432,
|
| 456 |
+
"grad_norm": 21.549283981323242,
|
| 457 |
+
"learning_rate": 5.471843634442683e-05,
|
| 458 |
+
"loss": 1.770701904296875,
|
| 459 |
+
"step": 5000
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.8804754567466432,
|
| 463 |
+
"eval_loss": 0.37507033348083496,
|
| 464 |
+
"eval_mIoU": 33.710236086968045,
|
| 465 |
+
"eval_mean_accuracy": 42.77628595940222,
|
| 466 |
+
"eval_overall_accuracy": 94.85613901377756,
|
| 467 |
+
"eval_runtime": 600.8517,
|
| 468 |
+
"eval_samples_per_second": 1.922,
|
| 469 |
+
"eval_steps_per_second": 0.962,
|
| 470 |
+
"step": 5000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.898084965881576,
|
| 474 |
+
"grad_norm": 4.752756595611572,
|
| 475 |
+
"learning_rate": 5.461278394083466e-05,
|
| 476 |
+
"loss": 1.8512596130371093,
|
| 477 |
+
"step": 5100
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.9156944750165089,
|
| 481 |
+
"grad_norm": 16.470531463623047,
|
| 482 |
+
"learning_rate": 5.4507131537242475e-05,
|
| 483 |
+
"loss": 1.6076771545410156,
|
| 484 |
+
"step": 5200
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 0.9333039841514418,
|
| 488 |
+
"grad_norm": 44.30925369262695,
|
| 489 |
+
"learning_rate": 5.440147913365029e-05,
|
| 490 |
+
"loss": 1.6375923156738281,
|
| 491 |
+
"step": 5300
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.9509134932863746,
|
| 495 |
+
"grad_norm": 13.600090980529785,
|
| 496 |
+
"learning_rate": 5.429582673005811e-05,
|
| 497 |
+
"loss": 1.5879168701171875,
|
| 498 |
+
"step": 5400
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.9685230024213075,
|
| 502 |
+
"grad_norm": 14.923215866088867,
|
| 503 |
+
"learning_rate": 5.419017432646593e-05,
|
| 504 |
+
"loss": 1.786272735595703,
|
| 505 |
+
"step": 5500
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.9685230024213075,
|
| 509 |
+
"eval_loss": 0.3556274175643921,
|
| 510 |
+
"eval_mIoU": 35.76118355021881,
|
| 511 |
+
"eval_mean_accuracy": 45.60329420506218,
|
| 512 |
+
"eval_overall_accuracy": 95.1163135875355,
|
| 513 |
+
"eval_runtime": 610.4692,
|
| 514 |
+
"eval_samples_per_second": 1.892,
|
| 515 |
+
"eval_steps_per_second": 0.947,
|
| 516 |
+
"step": 5500
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.9861325115562404,
|
| 520 |
+
"grad_norm": 13.075687408447266,
|
| 521 |
+
"learning_rate": 5.408452192287375e-05,
|
| 522 |
+
"loss": 1.6160453796386718,
|
| 523 |
+
"step": 5600
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 1.003697996918336,
|
| 527 |
+
"grad_norm": 11.696391105651855,
|
| 528 |
+
"learning_rate": 5.397886951928156e-05,
|
| 529 |
+
"loss": 1.6222117614746094,
|
| 530 |
+
"step": 5700
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 1.0213075060532688,
|
| 534 |
+
"grad_norm": 8.68483829498291,
|
| 535 |
+
"learning_rate": 5.3873217115689384e-05,
|
| 536 |
+
"loss": 1.2888230895996093,
|
| 537 |
+
"step": 5800
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 1.0389170151882017,
|
| 541 |
+
"grad_norm": 11.437637329101562,
|
| 542 |
+
"learning_rate": 5.37675647120972e-05,
|
| 543 |
+
"loss": 1.4501405334472657,
|
| 544 |
+
"step": 5900
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 1.0565265243231345,
|
| 548 |
+
"grad_norm": 23.864805221557617,
|
| 549 |
+
"learning_rate": 5.366191230850502e-05,
|
| 550 |
+
"loss": 1.4908659362792969,
|
| 551 |
+
"step": 6000
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 1.0565265243231345,
|
| 555 |
+
"eval_loss": 0.37875378131866455,
|
| 556 |
+
"eval_mIoU": 34.90903154363175,
|
| 557 |
+
"eval_mean_accuracy": 45.42566443381862,
|
| 558 |
+
"eval_overall_accuracy": 95.06103152320499,
|
| 559 |
+
"eval_runtime": 616.9187,
|
| 560 |
+
"eval_samples_per_second": 1.872,
|
| 561 |
+
"eval_steps_per_second": 0.937,
|
| 562 |
+
"step": 6000
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 1.0741360334580674,
|
| 566 |
+
"grad_norm": 5.937008857727051,
|
| 567 |
+
"learning_rate": 5.355625990491284e-05,
|
| 568 |
+
"loss": 1.3125173950195312,
|
| 569 |
+
"step": 6100
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 1.0917455425930003,
|
| 573 |
+
"grad_norm": 17.99025535583496,
|
| 574 |
+
"learning_rate": 5.3450607501320657e-05,
|
| 575 |
+
"loss": 1.5096702575683594,
|
| 576 |
+
"step": 6200
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 1.1093550517279331,
|
| 580 |
+
"grad_norm": 10.604606628417969,
|
| 581 |
+
"learning_rate": 5.3344955097728475e-05,
|
| 582 |
+
"loss": 1.3583757019042968,
|
| 583 |
+
"step": 6300
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 1.126964560862866,
|
| 587 |
+
"grad_norm": 11.066084861755371,
|
| 588 |
+
"learning_rate": 5.323930269413629e-05,
|
| 589 |
+
"loss": 1.4467390441894532,
|
| 590 |
+
"step": 6400
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 1.1445740699977989,
|
| 594 |
+
"grad_norm": 3.401655435562134,
|
| 595 |
+
"learning_rate": 5.313365029054411e-05,
|
| 596 |
+
"loss": 1.393478240966797,
|
| 597 |
+
"step": 6500
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 1.1445740699977989,
|
| 601 |
+
"eval_loss": 0.37657374143600464,
|
| 602 |
+
"eval_mIoU": 38.11100404001698,
|
| 603 |
+
"eval_mean_accuracy": 48.92294650624167,
|
| 604 |
+
"eval_overall_accuracy": 95.20334020837561,
|
| 605 |
+
"eval_runtime": 638.676,
|
| 606 |
+
"eval_samples_per_second": 1.808,
|
| 607 |
+
"eval_steps_per_second": 0.905,
|
| 608 |
+
"step": 6500
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 1.1621835791327317,
|
| 612 |
+
"grad_norm": 22.327383041381836,
|
| 613 |
+
"learning_rate": 5.302799788695193e-05,
|
| 614 |
+
"loss": 1.3139019775390626,
|
| 615 |
+
"step": 6600
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.1797930882676646,
|
| 619 |
+
"grad_norm": 25.93901252746582,
|
| 620 |
+
"learning_rate": 5.292234548335975e-05,
|
| 621 |
+
"loss": 1.4768368530273437,
|
| 622 |
+
"step": 6700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.1974025974025975,
|
| 626 |
+
"grad_norm": 11.60429573059082,
|
| 627 |
+
"learning_rate": 5.2816693079767565e-05,
|
| 628 |
+
"loss": 1.3625910949707032,
|
| 629 |
+
"step": 6800
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.2150121065375303,
|
| 633 |
+
"grad_norm": 4.614072799682617,
|
| 634 |
+
"learning_rate": 5.2711040676175384e-05,
|
| 635 |
+
"loss": 1.436598358154297,
|
| 636 |
+
"step": 6900
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.2326216156724632,
|
| 640 |
+
"grad_norm": 32.9177360534668,
|
| 641 |
+
"learning_rate": 5.26053882725832e-05,
|
| 642 |
+
"loss": 1.3963890075683594,
|
| 643 |
+
"step": 7000
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.2326216156724632,
|
| 647 |
+
"eval_loss": 0.36373773217201233,
|
| 648 |
+
"eval_mIoU": 38.23142578696723,
|
| 649 |
+
"eval_mean_accuracy": 48.00428340236774,
|
| 650 |
+
"eval_overall_accuracy": 95.36992324895157,
|
| 651 |
+
"eval_runtime": 623.8877,
|
| 652 |
+
"eval_samples_per_second": 1.851,
|
| 653 |
+
"eval_steps_per_second": 0.926,
|
| 654 |
+
"step": 7000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 1.250231124807396,
|
| 658 |
+
"grad_norm": 9.769020080566406,
|
| 659 |
+
"learning_rate": 5.249973586899103e-05,
|
| 660 |
+
"loss": 1.2838912963867188,
|
| 661 |
+
"step": 7100
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 1.267840633942329,
|
| 665 |
+
"grad_norm": 11.216667175292969,
|
| 666 |
+
"learning_rate": 5.239408346539884e-05,
|
| 667 |
+
"loss": 1.3618894958496093,
|
| 668 |
+
"step": 7200
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 1.2854501430772618,
|
| 672 |
+
"grad_norm": 3.4321463108062744,
|
| 673 |
+
"learning_rate": 5.2288431061806656e-05,
|
| 674 |
+
"loss": 1.2915570068359374,
|
| 675 |
+
"step": 7300
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 1.3030596522121947,
|
| 679 |
+
"grad_norm": 7.617103099822998,
|
| 680 |
+
"learning_rate": 5.2182778658214474e-05,
|
| 681 |
+
"loss": 1.3785501098632813,
|
| 682 |
+
"step": 7400
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 1.3206691613471275,
|
| 686 |
+
"grad_norm": 5.255311965942383,
|
| 687 |
+
"learning_rate": 5.207712625462229e-05,
|
| 688 |
+
"loss": 1.4688572692871094,
|
| 689 |
+
"step": 7500
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 1.3206691613471275,
|
| 693 |
+
"eval_loss": 0.3581092953681946,
|
| 694 |
+
"eval_mIoU": 36.78684041274723,
|
| 695 |
+
"eval_mean_accuracy": 46.12063614366271,
|
| 696 |
+
"eval_overall_accuracy": 95.14375926096203,
|
| 697 |
+
"eval_runtime": 632.8318,
|
| 698 |
+
"eval_samples_per_second": 1.825,
|
| 699 |
+
"eval_steps_per_second": 0.913,
|
| 700 |
+
"step": 7500
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 1.3382786704820604,
|
| 704 |
+
"grad_norm": 59.6718635559082,
|
| 705 |
+
"learning_rate": 5.197147385103011e-05,
|
| 706 |
+
"loss": 1.5980494689941407,
|
| 707 |
+
"step": 7600
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 1.355888179616993,
|
| 711 |
+
"grad_norm": 40.1579475402832,
|
| 712 |
+
"learning_rate": 5.186582144743793e-05,
|
| 713 |
+
"loss": 1.398350830078125,
|
| 714 |
+
"step": 7700
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.3734976887519261,
|
| 718 |
+
"grad_norm": 44.09864807128906,
|
| 719 |
+
"learning_rate": 5.1760169043845754e-05,
|
| 720 |
+
"loss": 1.4520651245117187,
|
| 721 |
+
"step": 7800
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.3911071978868588,
|
| 725 |
+
"grad_norm": 7.595520496368408,
|
| 726 |
+
"learning_rate": 5.1654516640253565e-05,
|
| 727 |
+
"loss": 1.414281005859375,
|
| 728 |
+
"step": 7900
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.4087167070217919,
|
| 732 |
+
"grad_norm": 22.861289978027344,
|
| 733 |
+
"learning_rate": 5.154886423666139e-05,
|
| 734 |
+
"loss": 1.3919259643554687,
|
| 735 |
+
"step": 8000
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 1.4087167070217919,
|
| 739 |
+
"eval_loss": 0.3777351677417755,
|
| 740 |
+
"eval_mIoU": 37.04354406013157,
|
| 741 |
+
"eval_mean_accuracy": 49.01216587877698,
|
| 742 |
+
"eval_overall_accuracy": 95.28077327843867,
|
| 743 |
+
"eval_runtime": 633.5258,
|
| 744 |
+
"eval_samples_per_second": 1.823,
|
| 745 |
+
"eval_steps_per_second": 0.912,
|
| 746 |
+
"step": 8000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.4263262161567245,
|
| 750 |
+
"grad_norm": 17.126745223999023,
|
| 751 |
+
"learning_rate": 5.14432118330692e-05,
|
| 752 |
+
"loss": 1.3992294311523437,
|
| 753 |
+
"step": 8100
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 1.4439357252916576,
|
| 757 |
+
"grad_norm": 25.60263442993164,
|
| 758 |
+
"learning_rate": 5.133755942947702e-05,
|
| 759 |
+
"loss": 1.5045437622070312,
|
| 760 |
+
"step": 8200
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 1.4615452344265902,
|
| 764 |
+
"grad_norm": 39.46979904174805,
|
| 765 |
+
"learning_rate": 5.123190702588484e-05,
|
| 766 |
+
"loss": 1.386280517578125,
|
| 767 |
+
"step": 8300
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.4791547435615233,
|
| 771 |
+
"grad_norm": 26.430803298950195,
|
| 772 |
+
"learning_rate": 5.1126254622292656e-05,
|
| 773 |
+
"loss": 1.395216827392578,
|
| 774 |
+
"step": 8400
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.496764252696456,
|
| 778 |
+
"grad_norm": 3.6278374195098877,
|
| 779 |
+
"learning_rate": 5.102060221870048e-05,
|
| 780 |
+
"loss": 1.3238540649414063,
|
| 781 |
+
"step": 8500
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 1.496764252696456,
|
| 785 |
+
"eval_loss": 0.3692784607410431,
|
| 786 |
+
"eval_mIoU": 38.03621070175311,
|
| 787 |
+
"eval_mean_accuracy": 49.83983428993074,
|
| 788 |
+
"eval_overall_accuracy": 95.30611112520292,
|
| 789 |
+
"eval_runtime": 632.3728,
|
| 790 |
+
"eval_samples_per_second": 1.826,
|
| 791 |
+
"eval_steps_per_second": 0.914,
|
| 792 |
+
"step": 8500
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 1.514373761831389,
|
| 796 |
+
"grad_norm": 10.106583595275879,
|
| 797 |
+
"learning_rate": 5.091494981510829e-05,
|
| 798 |
+
"loss": 1.341611328125,
|
| 799 |
+
"step": 8600
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 1.5319832709663217,
|
| 803 |
+
"grad_norm": 8.109859466552734,
|
| 804 |
+
"learning_rate": 5.080929741151612e-05,
|
| 805 |
+
"loss": 1.286796875,
|
| 806 |
+
"step": 8700
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 1.5495927801012548,
|
| 810 |
+
"grad_norm": 40.51174545288086,
|
| 811 |
+
"learning_rate": 5.070364500792393e-05,
|
| 812 |
+
"loss": 1.4733267211914063,
|
| 813 |
+
"step": 8800
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 1.5672022892361874,
|
| 817 |
+
"grad_norm": 8.529606819152832,
|
| 818 |
+
"learning_rate": 5.0597992604331754e-05,
|
| 819 |
+
"loss": 1.3715184020996094,
|
| 820 |
+
"step": 8900
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 1.5848117983711205,
|
| 824 |
+
"grad_norm": 10.91648006439209,
|
| 825 |
+
"learning_rate": 5.0492340200739565e-05,
|
| 826 |
+
"loss": 1.4365826416015626,
|
| 827 |
+
"step": 9000
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 1.5848117983711205,
|
| 831 |
+
"eval_loss": 0.3527390956878662,
|
| 832 |
+
"eval_mIoU": 38.72966019649981,
|
| 833 |
+
"eval_mean_accuracy": 50.33583101361378,
|
| 834 |
+
"eval_overall_accuracy": 95.30509519370604,
|
| 835 |
+
"eval_runtime": 622.2524,
|
| 836 |
+
"eval_samples_per_second": 1.856,
|
| 837 |
+
"eval_steps_per_second": 0.929,
|
| 838 |
+
"step": 9000
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.6024213075060532,
|
| 842 |
+
"grad_norm": 17.285547256469727,
|
| 843 |
+
"learning_rate": 5.038668779714738e-05,
|
| 844 |
+
"loss": 1.298778076171875,
|
| 845 |
+
"step": 9100
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.6200308166409862,
|
| 849 |
+
"grad_norm": 3.4417006969451904,
|
| 850 |
+
"learning_rate": 5.028103539355521e-05,
|
| 851 |
+
"loss": 1.3848799133300782,
|
| 852 |
+
"step": 9200
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.637640325775919,
|
| 856 |
+
"grad_norm": 11.60331916809082,
|
| 857 |
+
"learning_rate": 5.017538298996302e-05,
|
| 858 |
+
"loss": 1.396776123046875,
|
| 859 |
+
"step": 9300
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.655249834910852,
|
| 863 |
+
"grad_norm": 42.88685607910156,
|
| 864 |
+
"learning_rate": 5.0069730586370845e-05,
|
| 865 |
+
"loss": 1.5503982543945312,
|
| 866 |
+
"step": 9400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.6728593440457846,
|
| 870 |
+
"grad_norm": 10.325550079345703,
|
| 871 |
+
"learning_rate": 4.9964078182778656e-05,
|
| 872 |
+
"loss": 1.3557667541503906,
|
| 873 |
+
"step": 9500
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.6728593440457846,
|
| 877 |
+
"eval_loss": 0.3433685302734375,
|
| 878 |
+
"eval_mIoU": 37.68474411400127,
|
| 879 |
+
"eval_mean_accuracy": 48.22518380563681,
|
| 880 |
+
"eval_overall_accuracy": 95.38149978175308,
|
| 881 |
+
"eval_runtime": 643.0868,
|
| 882 |
+
"eval_samples_per_second": 1.796,
|
| 883 |
+
"eval_steps_per_second": 0.899,
|
| 884 |
+
"step": 9500
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.6904688531807177,
|
| 888 |
+
"grad_norm": 7.645025253295898,
|
| 889 |
+
"learning_rate": 4.985842577918648e-05,
|
| 890 |
+
"loss": 1.3318984985351563,
|
| 891 |
+
"step": 9600
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.7080783623156504,
|
| 895 |
+
"grad_norm": 2.253293514251709,
|
| 896 |
+
"learning_rate": 4.97527733755943e-05,
|
| 897 |
+
"loss": 1.4566912841796875,
|
| 898 |
+
"step": 9700
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 1.7256878714505834,
|
| 902 |
+
"grad_norm": 2.5841846466064453,
|
| 903 |
+
"learning_rate": 4.964712097200211e-05,
|
| 904 |
+
"loss": 1.3018017578125,
|
| 905 |
+
"step": 9800
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 1.743297380585516,
|
| 909 |
+
"grad_norm": 5.959505081176758,
|
| 910 |
+
"learning_rate": 4.9541468568409935e-05,
|
| 911 |
+
"loss": 1.3685440063476562,
|
| 912 |
+
"step": 9900
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.7609068897204492,
|
| 916 |
+
"grad_norm": 21.39006996154785,
|
| 917 |
+
"learning_rate": 4.943581616481775e-05,
|
| 918 |
+
"loss": 1.267444839477539,
|
| 919 |
+
"step": 10000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.7609068897204492,
|
| 923 |
+
"eval_loss": 0.3590288758277893,
|
| 924 |
+
"eval_mIoU": 39.250987764136376,
|
| 925 |
+
"eval_mean_accuracy": 51.270012242428145,
|
| 926 |
+
"eval_overall_accuracy": 95.32280232483174,
|
| 927 |
+
"eval_runtime": 643.0953,
|
| 928 |
+
"eval_samples_per_second": 1.796,
|
| 929 |
+
"eval_steps_per_second": 0.899,
|
| 930 |
+
"step": 10000
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.7785163988553818,
|
| 934 |
+
"grad_norm": 64.21682739257812,
|
| 935 |
+
"learning_rate": 4.933016376122557e-05,
|
| 936 |
+
"loss": 1.2918376159667968,
|
| 937 |
+
"step": 10100
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 1.796125907990315,
|
| 941 |
+
"grad_norm": 32.8283805847168,
|
| 942 |
+
"learning_rate": 4.922451135763338e-05,
|
| 943 |
+
"loss": 1.4452650451660156,
|
| 944 |
+
"step": 10200
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 1.8137354171252476,
|
| 948 |
+
"grad_norm": 11.545291900634766,
|
| 949 |
+
"learning_rate": 4.911885895404121e-05,
|
| 950 |
+
"loss": 1.2878465270996093,
|
| 951 |
+
"step": 10300
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 1.8313449262601806,
|
| 955 |
+
"grad_norm": 18.071176528930664,
|
| 956 |
+
"learning_rate": 4.9013206550449026e-05,
|
| 957 |
+
"loss": 1.3438943481445313,
|
| 958 |
+
"step": 10400
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 1.8489544353951133,
|
| 962 |
+
"grad_norm": 7.273808479309082,
|
| 963 |
+
"learning_rate": 4.8907554146856844e-05,
|
| 964 |
+
"loss": 1.2591412353515625,
|
| 965 |
+
"step": 10500
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 1.8489544353951133,
|
| 969 |
+
"eval_loss": 0.39277809858322144,
|
| 970 |
+
"eval_mIoU": 36.1147332767591,
|
| 971 |
+
"eval_mean_accuracy": 45.97273320824087,
|
| 972 |
+
"eval_overall_accuracy": 95.28447403020157,
|
| 973 |
+
"eval_runtime": 612.8948,
|
| 974 |
+
"eval_samples_per_second": 1.884,
|
| 975 |
+
"eval_steps_per_second": 0.943,
|
| 976 |
+
"step": 10500
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 1.8665639445300464,
|
| 980 |
+
"grad_norm": 27.94432830810547,
|
| 981 |
+
"learning_rate": 4.880190174326466e-05,
|
| 982 |
+
"loss": 1.218763427734375,
|
| 983 |
+
"step": 10600
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 1.884173453664979,
|
| 987 |
+
"grad_norm": 8.407777786254883,
|
| 988 |
+
"learning_rate": 4.8696249339672474e-05,
|
| 989 |
+
"loss": 1.3651194763183594,
|
| 990 |
+
"step": 10700
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 1.901782962799912,
|
| 994 |
+
"grad_norm": 20.770652770996094,
|
| 995 |
+
"learning_rate": 4.85905969360803e-05,
|
| 996 |
+
"loss": 1.3555903625488281,
|
| 997 |
+
"step": 10800
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 1.9193924719348447,
|
| 1001 |
+
"grad_norm": 13.678817749023438,
|
| 1002 |
+
"learning_rate": 4.848494453248811e-05,
|
| 1003 |
+
"loss": 1.3331718444824219,
|
| 1004 |
+
"step": 10900
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 1.9370019810697778,
|
| 1008 |
+
"grad_norm": 7.2792181968688965,
|
| 1009 |
+
"learning_rate": 4.8379292128895935e-05,
|
| 1010 |
+
"loss": 1.1297405242919922,
|
| 1011 |
+
"step": 11000
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 1.9370019810697778,
|
| 1015 |
+
"eval_loss": 0.36189690232276917,
|
| 1016 |
+
"eval_mIoU": 38.97650392834806,
|
| 1017 |
+
"eval_mean_accuracy": 50.780002728102545,
|
| 1018 |
+
"eval_overall_accuracy": 95.34987445517535,
|
| 1019 |
+
"eval_runtime": 612.1048,
|
| 1020 |
+
"eval_samples_per_second": 1.887,
|
| 1021 |
+
"eval_steps_per_second": 0.944,
|
| 1022 |
+
"step": 11000
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.9546114902047105,
|
| 1026 |
+
"grad_norm": 31.04173469543457,
|
| 1027 |
+
"learning_rate": 4.827363972530375e-05,
|
| 1028 |
+
"loss": 1.4935745239257812,
|
| 1029 |
+
"step": 11100
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.9722209993396436,
|
| 1033 |
+
"grad_norm": 20.730045318603516,
|
| 1034 |
+
"learning_rate": 4.816798732171157e-05,
|
| 1035 |
+
"loss": 1.4070867919921874,
|
| 1036 |
+
"step": 11200
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 1.9898305084745762,
|
| 1040 |
+
"grad_norm": 18.578184127807617,
|
| 1041 |
+
"learning_rate": 4.806233491811939e-05,
|
| 1042 |
+
"loss": 1.2718235778808593,
|
| 1043 |
+
"step": 11300
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 2.007395993836672,
|
| 1047 |
+
"grad_norm": 7.441476821899414,
|
| 1048 |
+
"learning_rate": 4.79566825145272e-05,
|
| 1049 |
+
"loss": 1.2224794769287108,
|
| 1050 |
+
"step": 11400
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 2.0250055029716045,
|
| 1054 |
+
"grad_norm": 7.709177494049072,
|
| 1055 |
+
"learning_rate": 4.7851030110935026e-05,
|
| 1056 |
+
"loss": 1.1453880310058593,
|
| 1057 |
+
"step": 11500
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 2.0250055029716045,
|
| 1061 |
+
"eval_loss": 0.3579792082309723,
|
| 1062 |
+
"eval_mIoU": 39.1309522428395,
|
| 1063 |
+
"eval_mean_accuracy": 49.93403955617436,
|
| 1064 |
+
"eval_overall_accuracy": 95.3825044838381,
|
| 1065 |
+
"eval_runtime": 615.2836,
|
| 1066 |
+
"eval_samples_per_second": 1.877,
|
| 1067 |
+
"eval_steps_per_second": 0.939,
|
| 1068 |
+
"step": 11500
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 2.0426150121065376,
|
| 1072 |
+
"grad_norm": 42.05501174926758,
|
| 1073 |
+
"learning_rate": 4.7745377707342844e-05,
|
| 1074 |
+
"loss": 1.2940916442871093,
|
| 1075 |
+
"step": 11600
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 2.0602245212414703,
|
| 1079 |
+
"grad_norm": 13.565258026123047,
|
| 1080 |
+
"learning_rate": 4.763972530375066e-05,
|
| 1081 |
+
"loss": 1.138370361328125,
|
| 1082 |
+
"step": 11700
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 2.0778340303764034,
|
| 1086 |
+
"grad_norm": 35.50266647338867,
|
| 1087 |
+
"learning_rate": 4.753407290015848e-05,
|
| 1088 |
+
"loss": 1.0196060180664062,
|
| 1089 |
+
"step": 11800
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 2.095443539511336,
|
| 1093 |
+
"grad_norm": 19.02262306213379,
|
| 1094 |
+
"learning_rate": 4.74284204965663e-05,
|
| 1095 |
+
"loss": 1.1524028778076172,
|
| 1096 |
+
"step": 11900
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 2.113053048646269,
|
| 1100 |
+
"grad_norm": 7.430933475494385,
|
| 1101 |
+
"learning_rate": 4.732276809297412e-05,
|
| 1102 |
+
"loss": 1.2204737854003906,
|
| 1103 |
+
"step": 12000
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 2.113053048646269,
|
| 1107 |
+
"eval_loss": 0.38723915815353394,
|
| 1108 |
+
"eval_mIoU": 39.744055139390696,
|
| 1109 |
+
"eval_mean_accuracy": 52.037140691988846,
|
| 1110 |
+
"eval_overall_accuracy": 95.38114407361844,
|
| 1111 |
+
"eval_runtime": 614.3063,
|
| 1112 |
+
"eval_samples_per_second": 1.88,
|
| 1113 |
+
"eval_steps_per_second": 0.941,
|
| 1114 |
+
"step": 12000
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 2.1306625577812017,
|
| 1118 |
+
"grad_norm": 12.622140884399414,
|
| 1119 |
+
"learning_rate": 4.7217115689381935e-05,
|
| 1120 |
+
"loss": 1.0923126220703125,
|
| 1121 |
+
"step": 12100
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 2.148272066916135,
|
| 1125 |
+
"grad_norm": 26.150724411010742,
|
| 1126 |
+
"learning_rate": 4.711146328578975e-05,
|
| 1127 |
+
"loss": 1.1749604797363282,
|
| 1128 |
+
"step": 12200
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 2.1658815760510675,
|
| 1132 |
+
"grad_norm": 12.140750885009766,
|
| 1133 |
+
"learning_rate": 4.700581088219757e-05,
|
| 1134 |
+
"loss": 1.1458908081054688,
|
| 1135 |
+
"step": 12300
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 2.1834910851860005,
|
| 1139 |
+
"grad_norm": 5.676041126251221,
|
| 1140 |
+
"learning_rate": 4.690015847860539e-05,
|
| 1141 |
+
"loss": 1.1439741516113282,
|
| 1142 |
+
"step": 12400
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 2.201100594320933,
|
| 1146 |
+
"grad_norm": 5.724509239196777,
|
| 1147 |
+
"learning_rate": 4.679450607501321e-05,
|
| 1148 |
+
"loss": 1.1061095428466796,
|
| 1149 |
+
"step": 12500
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 2.201100594320933,
|
| 1153 |
+
"eval_loss": 0.3751870095729828,
|
| 1154 |
+
"eval_mIoU": 39.40716115137035,
|
| 1155 |
+
"eval_mean_accuracy": 50.12508859035146,
|
| 1156 |
+
"eval_overall_accuracy": 95.35122363598316,
|
| 1157 |
+
"eval_runtime": 613.3143,
|
| 1158 |
+
"eval_samples_per_second": 1.883,
|
| 1159 |
+
"eval_steps_per_second": 0.942,
|
| 1160 |
+
"step": 12500
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 2.2187101034558663,
|
| 1164 |
+
"grad_norm": 21.661672592163086,
|
| 1165 |
+
"learning_rate": 4.6688853671421026e-05,
|
| 1166 |
+
"loss": 1.1017567443847656,
|
| 1167 |
+
"step": 12600
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 2.236319612590799,
|
| 1171 |
+
"grad_norm": 3.0723938941955566,
|
| 1172 |
+
"learning_rate": 4.6583201267828844e-05,
|
| 1173 |
+
"loss": 1.097019500732422,
|
| 1174 |
+
"step": 12700
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 2.253929121725732,
|
| 1178 |
+
"grad_norm": 20.003185272216797,
|
| 1179 |
+
"learning_rate": 4.647754886423666e-05,
|
| 1180 |
+
"loss": 1.0861209106445313,
|
| 1181 |
+
"step": 12800
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 2.2715386308606647,
|
| 1185 |
+
"grad_norm": 17.646045684814453,
|
| 1186 |
+
"learning_rate": 4.637189646064448e-05,
|
| 1187 |
+
"loss": 1.1462979888916016,
|
| 1188 |
+
"step": 12900
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 2.2891481399955977,
|
| 1192 |
+
"grad_norm": 6.879117488861084,
|
| 1193 |
+
"learning_rate": 4.62662440570523e-05,
|
| 1194 |
+
"loss": 1.2343074035644532,
|
| 1195 |
+
"step": 13000
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 2.2891481399955977,
|
| 1199 |
+
"eval_loss": 0.35330045223236084,
|
| 1200 |
+
"eval_mIoU": 39.403422282692816,
|
| 1201 |
+
"eval_mean_accuracy": 50.84406228069122,
|
| 1202 |
+
"eval_overall_accuracy": 95.43958259351326,
|
| 1203 |
+
"eval_runtime": 614.1751,
|
| 1204 |
+
"eval_samples_per_second": 1.881,
|
| 1205 |
+
"eval_steps_per_second": 0.941,
|
| 1206 |
+
"step": 13000
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 2.3067576491305304,
|
| 1210 |
+
"grad_norm": 83.66311645507812,
|
| 1211 |
+
"learning_rate": 4.616059165346012e-05,
|
| 1212 |
+
"loss": 1.0837049102783203,
|
| 1213 |
+
"step": 13100
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 2.3243671582654635,
|
| 1217 |
+
"grad_norm": 27.105422973632812,
|
| 1218 |
+
"learning_rate": 4.6054939249867935e-05,
|
| 1219 |
+
"loss": 1.2298228454589843,
|
| 1220 |
+
"step": 13200
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 2.341976667400396,
|
| 1224 |
+
"grad_norm": 7.361494541168213,
|
| 1225 |
+
"learning_rate": 4.594928684627575e-05,
|
| 1226 |
+
"loss": 1.1874884796142577,
|
| 1227 |
+
"step": 13300
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 2.359586176535329,
|
| 1231 |
+
"grad_norm": 5.626405239105225,
|
| 1232 |
+
"learning_rate": 4.584363444268357e-05,
|
| 1233 |
+
"loss": 1.1711844635009765,
|
| 1234 |
+
"step": 13400
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 2.377195685670262,
|
| 1238 |
+
"grad_norm": 5.629500865936279,
|
| 1239 |
+
"learning_rate": 4.5737982039091396e-05,
|
| 1240 |
+
"loss": 1.1417552947998046,
|
| 1241 |
+
"step": 13500
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 2.377195685670262,
|
| 1245 |
+
"eval_loss": 0.3440675735473633,
|
| 1246 |
+
"eval_mIoU": 41.19118866588967,
|
| 1247 |
+
"eval_mean_accuracy": 52.981668891642265,
|
| 1248 |
+
"eval_overall_accuracy": 95.56694691315359,
|
| 1249 |
+
"eval_runtime": 615.9929,
|
| 1250 |
+
"eval_samples_per_second": 1.875,
|
| 1251 |
+
"eval_steps_per_second": 0.938,
|
| 1252 |
+
"step": 13500
|
| 1253 |
+
},
|
| 1254 |
+
{
|
| 1255 |
+
"epoch": 2.394805194805195,
|
| 1256 |
+
"grad_norm": 7.499139785766602,
|
| 1257 |
+
"learning_rate": 4.563232963549921e-05,
|
| 1258 |
+
"loss": 1.1351351165771484,
|
| 1259 |
+
"step": 13600
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 2.4124147039401276,
|
| 1263 |
+
"grad_norm": 6.671785831451416,
|
| 1264 |
+
"learning_rate": 4.552667723190703e-05,
|
| 1265 |
+
"loss": 1.158738479614258,
|
| 1266 |
+
"step": 13700
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 2.4300242130750607,
|
| 1270 |
+
"grad_norm": 5.020116806030273,
|
| 1271 |
+
"learning_rate": 4.5421024828314844e-05,
|
| 1272 |
+
"loss": 1.0873625183105469,
|
| 1273 |
+
"step": 13800
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 2.4476337222099933,
|
| 1277 |
+
"grad_norm": 11.06270694732666,
|
| 1278 |
+
"learning_rate": 4.531537242472266e-05,
|
| 1279 |
+
"loss": 1.1327620697021485,
|
| 1280 |
+
"step": 13900
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 2.4652432313449264,
|
| 1284 |
+
"grad_norm": 39.09502410888672,
|
| 1285 |
+
"learning_rate": 4.520972002113048e-05,
|
| 1286 |
+
"loss": 1.163224563598633,
|
| 1287 |
+
"step": 14000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 2.4652432313449264,
|
| 1291 |
+
"eval_loss": 0.37537702918052673,
|
| 1292 |
+
"eval_mIoU": 37.81560249206644,
|
| 1293 |
+
"eval_mean_accuracy": 47.31413884036508,
|
| 1294 |
+
"eval_overall_accuracy": 95.4164120232388,
|
| 1295 |
+
"eval_runtime": 621.4201,
|
| 1296 |
+
"eval_samples_per_second": 1.859,
|
| 1297 |
+
"eval_steps_per_second": 0.93,
|
| 1298 |
+
"step": 14000
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.482852740479859,
|
| 1302 |
+
"grad_norm": 3.848895311355591,
|
| 1303 |
+
"learning_rate": 4.51040676175383e-05,
|
| 1304 |
+
"loss": 1.0415728759765626,
|
| 1305 |
+
"step": 14100
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.500462249614792,
|
| 1309 |
+
"grad_norm": 4.380424976348877,
|
| 1310 |
+
"learning_rate": 4.499841521394612e-05,
|
| 1311 |
+
"loss": 1.1635037994384765,
|
| 1312 |
+
"step": 14200
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 2.5180717587497248,
|
| 1316 |
+
"grad_norm": 37.84204864501953,
|
| 1317 |
+
"learning_rate": 4.4892762810353935e-05,
|
| 1318 |
+
"loss": 1.2334834289550782,
|
| 1319 |
+
"step": 14300
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.535681267884658,
|
| 1323 |
+
"grad_norm": 7.818023204803467,
|
| 1324 |
+
"learning_rate": 4.478711040676176e-05,
|
| 1325 |
+
"loss": 1.0910034942626954,
|
| 1326 |
+
"step": 14400
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 2.5532907770195905,
|
| 1330 |
+
"grad_norm": 19.556774139404297,
|
| 1331 |
+
"learning_rate": 4.468145800316957e-05,
|
| 1332 |
+
"loss": 1.1642417907714844,
|
| 1333 |
+
"step": 14500
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 2.5532907770195905,
|
| 1337 |
+
"eval_loss": 0.3476269245147705,
|
| 1338 |
+
"eval_mIoU": 39.42816513930243,
|
| 1339 |
+
"eval_mean_accuracy": 50.40911446506301,
|
| 1340 |
+
"eval_overall_accuracy": 95.54313032142011,
|
| 1341 |
+
"eval_runtime": 630.7714,
|
| 1342 |
+
"eval_samples_per_second": 1.831,
|
| 1343 |
+
"eval_steps_per_second": 0.916,
|
| 1344 |
+
"step": 14500
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 2.5709002861545236,
|
| 1348 |
+
"grad_norm": 38.74016189575195,
|
| 1349 |
+
"learning_rate": 4.457580559957739e-05,
|
| 1350 |
+
"loss": 1.0831634521484375,
|
| 1351 |
+
"step": 14600
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 2.5885097952894562,
|
| 1355 |
+
"grad_norm": 11.159551620483398,
|
| 1356 |
+
"learning_rate": 4.447015319598521e-05,
|
| 1357 |
+
"loss": 1.1328179168701171,
|
| 1358 |
+
"step": 14700
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 2.6061193044243893,
|
| 1362 |
+
"grad_norm": 7.955896854400635,
|
| 1363 |
+
"learning_rate": 4.4364500792393026e-05,
|
| 1364 |
+
"loss": 1.124356231689453,
|
| 1365 |
+
"step": 14800
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 2.623728813559322,
|
| 1369 |
+
"grad_norm": 6.735036849975586,
|
| 1370 |
+
"learning_rate": 4.425884838880085e-05,
|
| 1371 |
+
"loss": 1.060822982788086,
|
| 1372 |
+
"step": 14900
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 2.641338322694255,
|
| 1376 |
+
"grad_norm": 6.450618743896484,
|
| 1377 |
+
"learning_rate": 4.415319598520866e-05,
|
| 1378 |
+
"loss": 1.0960148620605468,
|
| 1379 |
+
"step": 15000
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 2.641338322694255,
|
| 1383 |
+
"eval_loss": 0.3715696632862091,
|
| 1384 |
+
"eval_mIoU": 40.637422462526686,
|
| 1385 |
+
"eval_mean_accuracy": 51.30163306486706,
|
| 1386 |
+
"eval_overall_accuracy": 95.44940700778713,
|
| 1387 |
+
"eval_runtime": 632.007,
|
| 1388 |
+
"eval_samples_per_second": 1.828,
|
| 1389 |
+
"eval_steps_per_second": 0.915,
|
| 1390 |
+
"step": 15000
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 2.6589478318291877,
|
| 1394 |
+
"grad_norm": 5.860879421234131,
|
| 1395 |
+
"learning_rate": 4.404754358161649e-05,
|
| 1396 |
+
"loss": 1.030592041015625,
|
| 1397 |
+
"step": 15100
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 2.676557340964121,
|
| 1401 |
+
"grad_norm": 5.151623725891113,
|
| 1402 |
+
"learning_rate": 4.39418911780243e-05,
|
| 1403 |
+
"loss": 0.9981183624267578,
|
| 1404 |
+
"step": 15200
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 2.6941668500990534,
|
| 1408 |
+
"grad_norm": 51.52006149291992,
|
| 1409 |
+
"learning_rate": 4.383623877443212e-05,
|
| 1410 |
+
"loss": 1.1631646728515626,
|
| 1411 |
+
"step": 15300
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 2.711776359233986,
|
| 1415 |
+
"grad_norm": 1.7929599285125732,
|
| 1416 |
+
"learning_rate": 4.3730586370839935e-05,
|
| 1417 |
+
"loss": 1.034713134765625,
|
| 1418 |
+
"step": 15400
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 2.729385868368919,
|
| 1422 |
+
"grad_norm": 79.84156799316406,
|
| 1423 |
+
"learning_rate": 4.362493396724775e-05,
|
| 1424 |
+
"loss": 1.1282410430908203,
|
| 1425 |
+
"step": 15500
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 2.729385868368919,
|
| 1429 |
+
"eval_loss": 0.34207868576049805,
|
| 1430 |
+
"eval_mIoU": 41.725837022101224,
|
| 1431 |
+
"eval_mean_accuracy": 51.67503102984655,
|
| 1432 |
+
"eval_overall_accuracy": 95.68064702021611,
|
| 1433 |
+
"eval_runtime": 646.8847,
|
| 1434 |
+
"eval_samples_per_second": 1.785,
|
| 1435 |
+
"eval_steps_per_second": 0.894,
|
| 1436 |
+
"step": 15500
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 2.7469953775038523,
|
| 1440 |
+
"grad_norm": 3.692952871322632,
|
| 1441 |
+
"learning_rate": 4.351928156365558e-05,
|
| 1442 |
+
"loss": 1.072099609375,
|
| 1443 |
+
"step": 15600
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 2.764604886638785,
|
| 1447 |
+
"grad_norm": 57.948753356933594,
|
| 1448 |
+
"learning_rate": 4.341362916006339e-05,
|
| 1449 |
+
"loss": 1.1195967864990235,
|
| 1450 |
+
"step": 15700
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 2.7822143957737175,
|
| 1454 |
+
"grad_norm": 8.935181617736816,
|
| 1455 |
+
"learning_rate": 4.3307976756471214e-05,
|
| 1456 |
+
"loss": 1.0987003326416016,
|
| 1457 |
+
"step": 15800
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 2.7998239049086506,
|
| 1461 |
+
"grad_norm": 3.0621280670166016,
|
| 1462 |
+
"learning_rate": 4.3202324352879025e-05,
|
| 1463 |
+
"loss": 1.1626800537109374,
|
| 1464 |
+
"step": 15900
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 2.8174334140435837,
|
| 1468 |
+
"grad_norm": 11.499585151672363,
|
| 1469 |
+
"learning_rate": 4.309667194928685e-05,
|
| 1470 |
+
"loss": 1.2044084930419923,
|
| 1471 |
+
"step": 16000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 2.8174334140435837,
|
| 1475 |
+
"eval_loss": 0.4090178608894348,
|
| 1476 |
+
"eval_mIoU": 38.788988881844546,
|
| 1477 |
+
"eval_mean_accuracy": 49.14422837445873,
|
| 1478 |
+
"eval_overall_accuracy": 95.46250347451213,
|
| 1479 |
+
"eval_runtime": 686.1121,
|
| 1480 |
+
"eval_samples_per_second": 1.683,
|
| 1481 |
+
"eval_steps_per_second": 0.842,
|
| 1482 |
+
"step": 16000
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 2.8350429231785164,
|
| 1486 |
+
"grad_norm": 7.883459091186523,
|
| 1487 |
+
"learning_rate": 4.299101954569467e-05,
|
| 1488 |
+
"loss": 1.1145830535888672,
|
| 1489 |
+
"step": 16100
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 2.852652432313449,
|
| 1493 |
+
"grad_norm": 26.763925552368164,
|
| 1494 |
+
"learning_rate": 4.288536714210248e-05,
|
| 1495 |
+
"loss": 1.1529721069335936,
|
| 1496 |
+
"step": 16200
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 2.870261941448382,
|
| 1500 |
+
"grad_norm": 11.193148612976074,
|
| 1501 |
+
"learning_rate": 4.2779714738510305e-05,
|
| 1502 |
+
"loss": 1.1887353515625,
|
| 1503 |
+
"step": 16300
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 2.887871450583315,
|
| 1507 |
+
"grad_norm": 47.4949836730957,
|
| 1508 |
+
"learning_rate": 4.2674062334918116e-05,
|
| 1509 |
+
"loss": 1.2108176422119141,
|
| 1510 |
+
"step": 16400
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 2.905480959718248,
|
| 1514 |
+
"grad_norm": 18.99823570251465,
|
| 1515 |
+
"learning_rate": 4.256840993132594e-05,
|
| 1516 |
+
"loss": 1.0943442535400392,
|
| 1517 |
+
"step": 16500
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 2.905480959718248,
|
| 1521 |
+
"eval_loss": 0.35304567217826843,
|
| 1522 |
+
"eval_mIoU": 41.13725855455199,
|
| 1523 |
+
"eval_mean_accuracy": 52.20547744211784,
|
| 1524 |
+
"eval_overall_accuracy": 95.69274142707064,
|
| 1525 |
+
"eval_runtime": 591.0741,
|
| 1526 |
+
"eval_samples_per_second": 1.954,
|
| 1527 |
+
"eval_steps_per_second": 0.978,
|
| 1528 |
+
"step": 16500
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 2.9230904688531805,
|
| 1532 |
+
"grad_norm": 6.136702537536621,
|
| 1533 |
+
"learning_rate": 4.246275752773375e-05,
|
| 1534 |
+
"loss": 1.0625466156005858,
|
| 1535 |
+
"step": 16600
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 2.9406999779881136,
|
| 1539 |
+
"grad_norm": 14.154541015625,
|
| 1540 |
+
"learning_rate": 4.235710512414158e-05,
|
| 1541 |
+
"loss": 1.0732533264160156,
|
| 1542 |
+
"step": 16700
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 2.9583094871230466,
|
| 1546 |
+
"grad_norm": 3.0931496620178223,
|
| 1547 |
+
"learning_rate": 4.2251452720549396e-05,
|
| 1548 |
+
"loss": 1.0319291687011718,
|
| 1549 |
+
"step": 16800
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 2.9759189962579793,
|
| 1553 |
+
"grad_norm": 80.85027313232422,
|
| 1554 |
+
"learning_rate": 4.2145800316957214e-05,
|
| 1555 |
+
"loss": 1.0832841491699219,
|
| 1556 |
+
"step": 16900
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 2.993528505392912,
|
| 1560 |
+
"grad_norm": 3.672075033187866,
|
| 1561 |
+
"learning_rate": 4.204014791336503e-05,
|
| 1562 |
+
"loss": 1.2637503051757812,
|
| 1563 |
+
"step": 17000
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 2.993528505392912,
|
| 1567 |
+
"eval_loss": 0.33654382824897766,
|
| 1568 |
+
"eval_mIoU": 41.44872246605363,
|
| 1569 |
+
"eval_mean_accuracy": 54.77046429937636,
|
| 1570 |
+
"eval_overall_accuracy": 95.56333863890016,
|
| 1571 |
+
"eval_runtime": 595.7763,
|
| 1572 |
+
"eval_samples_per_second": 1.939,
|
| 1573 |
+
"eval_steps_per_second": 0.97,
|
| 1574 |
+
"step": 17000
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 3.0110939907550076,
|
| 1578 |
+
"grad_norm": 7.291171550750732,
|
| 1579 |
+
"learning_rate": 4.193449550977284e-05,
|
| 1580 |
+
"loss": 1.0205813598632814,
|
| 1581 |
+
"step": 17100
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 3.0287034998899407,
|
| 1585 |
+
"grad_norm": 5.288455009460449,
|
| 1586 |
+
"learning_rate": 4.182884310618067e-05,
|
| 1587 |
+
"loss": 1.025837631225586,
|
| 1588 |
+
"step": 17200
|
| 1589 |
+
},
|
| 1590 |
+
{
|
| 1591 |
+
"epoch": 3.0463130090248733,
|
| 1592 |
+
"grad_norm": 2.849029302597046,
|
| 1593 |
+
"learning_rate": 4.172319070258848e-05,
|
| 1594 |
+
"loss": 0.8235783386230469,
|
| 1595 |
+
"step": 17300
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 3.0639225181598064,
|
| 1599 |
+
"grad_norm": 4.3493499755859375,
|
| 1600 |
+
"learning_rate": 4.1617538298996305e-05,
|
| 1601 |
+
"loss": 0.9443639373779297,
|
| 1602 |
+
"step": 17400
|
| 1603 |
+
},
|
| 1604 |
+
{
|
| 1605 |
+
"epoch": 3.081532027294739,
|
| 1606 |
+
"grad_norm": 17.059249877929688,
|
| 1607 |
+
"learning_rate": 4.151188589540412e-05,
|
| 1608 |
+
"loss": 0.9543634033203126,
|
| 1609 |
+
"step": 17500
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 3.081532027294739,
|
| 1613 |
+
"eval_loss": 0.3980509042739868,
|
| 1614 |
+
"eval_mIoU": 39.32169521997824,
|
| 1615 |
+
"eval_mean_accuracy": 48.75387080815462,
|
| 1616 |
+
"eval_overall_accuracy": 95.51444445853625,
|
| 1617 |
+
"eval_runtime": 595.8295,
|
| 1618 |
+
"eval_samples_per_second": 1.938,
|
| 1619 |
+
"eval_steps_per_second": 0.97,
|
| 1620 |
+
"step": 17500
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 3.099141536429672,
|
| 1624 |
+
"grad_norm": 20.158220291137695,
|
| 1625 |
+
"learning_rate": 4.140623349181194e-05,
|
| 1626 |
+
"loss": 0.96615478515625,
|
| 1627 |
+
"step": 17600
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 3.116751045564605,
|
| 1631 |
+
"grad_norm": 13.208799362182617,
|
| 1632 |
+
"learning_rate": 4.130058108821976e-05,
|
| 1633 |
+
"loss": 0.8107662963867187,
|
| 1634 |
+
"step": 17700
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 3.134360554699538,
|
| 1638 |
+
"grad_norm": 1.974561095237732,
|
| 1639 |
+
"learning_rate": 4.119492868462757e-05,
|
| 1640 |
+
"loss": 0.8889852142333985,
|
| 1641 |
+
"step": 17800
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 3.1519700638344705,
|
| 1645 |
+
"grad_norm": 19.499155044555664,
|
| 1646 |
+
"learning_rate": 4.1089276281035395e-05,
|
| 1647 |
+
"loss": 0.8983279418945312,
|
| 1648 |
+
"step": 17900
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 3.1695795729694036,
|
| 1652 |
+
"grad_norm": 7.43156623840332,
|
| 1653 |
+
"learning_rate": 4.0983623877443214e-05,
|
| 1654 |
+
"loss": 0.843604736328125,
|
| 1655 |
+
"step": 18000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 3.1695795729694036,
|
| 1659 |
+
"eval_loss": 0.3665158748626709,
|
| 1660 |
+
"eval_mIoU": 40.75046102280888,
|
| 1661 |
+
"eval_mean_accuracy": 51.73736939649708,
|
| 1662 |
+
"eval_overall_accuracy": 95.71766147365818,
|
| 1663 |
+
"eval_runtime": 597.7525,
|
| 1664 |
+
"eval_samples_per_second": 1.932,
|
| 1665 |
+
"eval_steps_per_second": 0.967,
|
| 1666 |
+
"step": 18000
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 3.1871890821043363,
|
| 1670 |
+
"grad_norm": 8.212807655334473,
|
| 1671 |
+
"learning_rate": 4.087797147385103e-05,
|
| 1672 |
+
"loss": 0.8693031311035156,
|
| 1673 |
+
"step": 18100
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 3.2047985912392694,
|
| 1677 |
+
"grad_norm": 10.272090911865234,
|
| 1678 |
+
"learning_rate": 4.077231907025885e-05,
|
| 1679 |
+
"loss": 0.7922476196289062,
|
| 1680 |
+
"step": 18200
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 3.222408100374202,
|
| 1684 |
+
"grad_norm": 9.316787719726562,
|
| 1685 |
+
"learning_rate": 4.066666666666667e-05,
|
| 1686 |
+
"loss": 0.9222467803955078,
|
| 1687 |
+
"step": 18300
|
| 1688 |
+
},
|
| 1689 |
+
{
|
| 1690 |
+
"epoch": 3.240017609509135,
|
| 1691 |
+
"grad_norm": 5.906984329223633,
|
| 1692 |
+
"learning_rate": 4.0561014263074486e-05,
|
| 1693 |
+
"loss": 0.933861312866211,
|
| 1694 |
+
"step": 18400
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 3.2576271186440677,
|
| 1698 |
+
"grad_norm": 35.925132751464844,
|
| 1699 |
+
"learning_rate": 4.0455361859482304e-05,
|
| 1700 |
+
"loss": 1.147016830444336,
|
| 1701 |
+
"step": 18500
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 3.2576271186440677,
|
| 1705 |
+
"eval_loss": 0.3637526333332062,
|
| 1706 |
+
"eval_mIoU": 40.01437254491868,
|
| 1707 |
+
"eval_mean_accuracy": 52.52752892592268,
|
| 1708 |
+
"eval_overall_accuracy": 95.57071008723535,
|
| 1709 |
+
"eval_runtime": 598.7356,
|
| 1710 |
+
"eval_samples_per_second": 1.929,
|
| 1711 |
+
"eval_steps_per_second": 0.965,
|
| 1712 |
+
"step": 18500
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 3.275236627779001,
|
| 1716 |
+
"grad_norm": 44.39639663696289,
|
| 1717 |
+
"learning_rate": 4.034970945589012e-05,
|
| 1718 |
+
"loss": 0.9909713745117188,
|
| 1719 |
+
"step": 18600
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 3.2928461369139335,
|
| 1723 |
+
"grad_norm": 7.367189407348633,
|
| 1724 |
+
"learning_rate": 4.024405705229794e-05,
|
| 1725 |
+
"loss": 0.899927978515625,
|
| 1726 |
+
"step": 18700
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 3.3104556460488666,
|
| 1730 |
+
"grad_norm": 3.2386536598205566,
|
| 1731 |
+
"learning_rate": 4.013840464870576e-05,
|
| 1732 |
+
"loss": 0.900641860961914,
|
| 1733 |
+
"step": 18800
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 3.328065155183799,
|
| 1737 |
+
"grad_norm": 5.339411735534668,
|
| 1738 |
+
"learning_rate": 4.003275224511358e-05,
|
| 1739 |
+
"loss": 0.8285768127441406,
|
| 1740 |
+
"step": 18900
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 3.3456746643187323,
|
| 1744 |
+
"grad_norm": 7.77389669418335,
|
| 1745 |
+
"learning_rate": 3.9927099841521395e-05,
|
| 1746 |
+
"loss": 1.016840362548828,
|
| 1747 |
+
"step": 19000
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 3.3456746643187323,
|
| 1751 |
+
"eval_loss": 0.35560256242752075,
|
| 1752 |
+
"eval_mIoU": 41.41963957275985,
|
| 1753 |
+
"eval_mean_accuracy": 52.9759184171816,
|
| 1754 |
+
"eval_overall_accuracy": 95.71307690112621,
|
| 1755 |
+
"eval_runtime": 598.8628,
|
| 1756 |
+
"eval_samples_per_second": 1.929,
|
| 1757 |
+
"eval_steps_per_second": 0.965,
|
| 1758 |
+
"step": 19000
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 3.363284173453665,
|
| 1762 |
+
"grad_norm": 13.394330024719238,
|
| 1763 |
+
"learning_rate": 3.9821447437929213e-05,
|
| 1764 |
+
"loss": 0.8661341857910156,
|
| 1765 |
+
"step": 19100
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 3.380893682588598,
|
| 1769 |
+
"grad_norm": 6.459191799163818,
|
| 1770 |
+
"learning_rate": 3.971579503433703e-05,
|
| 1771 |
+
"loss": 0.8622269439697265,
|
| 1772 |
+
"step": 19200
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 3.3985031917235307,
|
| 1776 |
+
"grad_norm": 16.69118309020996,
|
| 1777 |
+
"learning_rate": 3.961014263074485e-05,
|
| 1778 |
+
"loss": 0.9276141357421875,
|
| 1779 |
+
"step": 19300
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 3.4161127008584637,
|
| 1783 |
+
"grad_norm": 11.399658203125,
|
| 1784 |
+
"learning_rate": 3.9504490227152675e-05,
|
| 1785 |
+
"loss": 0.9722150421142578,
|
| 1786 |
+
"step": 19400
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 3.4337222099933964,
|
| 1790 |
+
"grad_norm": 14.789872169494629,
|
| 1791 |
+
"learning_rate": 3.9398837823560486e-05,
|
| 1792 |
+
"loss": 0.930595703125,
|
| 1793 |
+
"step": 19500
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 3.4337222099933964,
|
| 1797 |
+
"eval_loss": 0.3928566873073578,
|
| 1798 |
+
"eval_mIoU": 40.82311861205255,
|
| 1799 |
+
"eval_mean_accuracy": 51.636349706196874,
|
| 1800 |
+
"eval_overall_accuracy": 95.65218277307815,
|
| 1801 |
+
"eval_runtime": 599.9531,
|
| 1802 |
+
"eval_samples_per_second": 1.925,
|
| 1803 |
+
"eval_steps_per_second": 0.963,
|
| 1804 |
+
"step": 19500
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 3.4513317191283295,
|
| 1808 |
+
"grad_norm": 47.58243942260742,
|
| 1809 |
+
"learning_rate": 3.9293185419968304e-05,
|
| 1810 |
+
"loss": 0.948566665649414,
|
| 1811 |
+
"step": 19600
|
| 1812 |
+
},
|
| 1813 |
+
{
|
| 1814 |
+
"epoch": 3.468941228263262,
|
| 1815 |
+
"grad_norm": 10.375826835632324,
|
| 1816 |
+
"learning_rate": 3.918753301637612e-05,
|
| 1817 |
+
"loss": 0.9331759643554688,
|
| 1818 |
+
"step": 19700
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 3.486550737398195,
|
| 1822 |
+
"grad_norm": 27.57236099243164,
|
| 1823 |
+
"learning_rate": 3.908188061278394e-05,
|
| 1824 |
+
"loss": 0.9229088592529296,
|
| 1825 |
+
"step": 19800
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 3.504160246533128,
|
| 1829 |
+
"grad_norm": 14.850144386291504,
|
| 1830 |
+
"learning_rate": 3.897622820919176e-05,
|
| 1831 |
+
"loss": 0.9040148162841797,
|
| 1832 |
+
"step": 19900
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 3.5217697556680605,
|
| 1836 |
+
"grad_norm": 9.243043899536133,
|
| 1837 |
+
"learning_rate": 3.887057580559958e-05,
|
| 1838 |
+
"loss": 1.018120574951172,
|
| 1839 |
+
"step": 20000
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 3.5217697556680605,
|
| 1843 |
+
"eval_loss": 0.3454851508140564,
|
| 1844 |
+
"eval_mIoU": 41.75683968881324,
|
| 1845 |
+
"eval_mean_accuracy": 54.63857579379155,
|
| 1846 |
+
"eval_overall_accuracy": 95.76613058775534,
|
| 1847 |
+
"eval_runtime": 613.735,
|
| 1848 |
+
"eval_samples_per_second": 1.882,
|
| 1849 |
+
"eval_steps_per_second": 0.942,
|
| 1850 |
+
"step": 20000
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 3.5393792648029936,
|
| 1854 |
+
"grad_norm": 19.1420955657959,
|
| 1855 |
+
"learning_rate": 3.87649234020074e-05,
|
| 1856 |
+
"loss": 0.8436627197265625,
|
| 1857 |
+
"step": 20100
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 3.5569887739379267,
|
| 1861 |
+
"grad_norm": 15.668157577514648,
|
| 1862 |
+
"learning_rate": 3.865927099841521e-05,
|
| 1863 |
+
"loss": 0.9943603515625,
|
| 1864 |
+
"step": 20200
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 3.5745982830728593,
|
| 1868 |
+
"grad_norm": 8.847800254821777,
|
| 1869 |
+
"learning_rate": 3.855361859482303e-05,
|
| 1870 |
+
"loss": 0.959801254272461,
|
| 1871 |
+
"step": 20300
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 3.592207792207792,
|
| 1875 |
+
"grad_norm": 14.564043045043945,
|
| 1876 |
+
"learning_rate": 3.844796619123085e-05,
|
| 1877 |
+
"loss": 0.8551026153564453,
|
| 1878 |
+
"step": 20400
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 3.609817301342725,
|
| 1882 |
+
"grad_norm": 9.54574203491211,
|
| 1883 |
+
"learning_rate": 3.834231378763867e-05,
|
| 1884 |
+
"loss": 0.843503189086914,
|
| 1885 |
+
"step": 20500
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 3.609817301342725,
|
| 1889 |
+
"eval_loss": 0.3464469313621521,
|
| 1890 |
+
"eval_mIoU": 41.94745733642343,
|
| 1891 |
+
"eval_mean_accuracy": 51.44057812491615,
|
| 1892 |
+
"eval_overall_accuracy": 95.820711474398,
|
| 1893 |
+
"eval_runtime": 619.4723,
|
| 1894 |
+
"eval_samples_per_second": 1.864,
|
| 1895 |
+
"eval_steps_per_second": 0.933,
|
| 1896 |
+
"step": 20500
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 3.627426810477658,
|
| 1900 |
+
"grad_norm": 6.365943431854248,
|
| 1901 |
+
"learning_rate": 3.823666138404649e-05,
|
| 1902 |
+
"loss": 0.9280338287353516,
|
| 1903 |
+
"step": 20600
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 3.645036319612591,
|
| 1907 |
+
"grad_norm": 49.77190017700195,
|
| 1908 |
+
"learning_rate": 3.8131008980454304e-05,
|
| 1909 |
+
"loss": 0.814898681640625,
|
| 1910 |
+
"step": 20700
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 3.6626458287475234,
|
| 1914 |
+
"grad_norm": 22.503215789794922,
|
| 1915 |
+
"learning_rate": 3.802535657686213e-05,
|
| 1916 |
+
"loss": 0.9596369171142578,
|
| 1917 |
+
"step": 20800
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 3.6802553378824565,
|
| 1921 |
+
"grad_norm": 3.472125291824341,
|
| 1922 |
+
"learning_rate": 3.791970417326994e-05,
|
| 1923 |
+
"loss": 0.7934391021728515,
|
| 1924 |
+
"step": 20900
|
| 1925 |
+
},
|
| 1926 |
+
{
|
| 1927 |
+
"epoch": 3.6978648470173896,
|
| 1928 |
+
"grad_norm": 7.981621265411377,
|
| 1929 |
+
"learning_rate": 3.7814051769677765e-05,
|
| 1930 |
+
"loss": 0.9479257202148438,
|
| 1931 |
+
"step": 21000
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 3.6978648470173896,
|
| 1935 |
+
"eval_loss": 0.3813023269176483,
|
| 1936 |
+
"eval_mIoU": 40.63642183599385,
|
| 1937 |
+
"eval_mean_accuracy": 54.05427297237617,
|
| 1938 |
+
"eval_overall_accuracy": 95.5580978063278,
|
| 1939 |
+
"eval_runtime": 615.8201,
|
| 1940 |
+
"eval_samples_per_second": 1.876,
|
| 1941 |
+
"eval_steps_per_second": 0.939,
|
| 1942 |
+
"step": 21000
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 3.7154743561523222,
|
| 1946 |
+
"grad_norm": 28.353187561035156,
|
| 1947 |
+
"learning_rate": 3.770839936608558e-05,
|
| 1948 |
+
"loss": 0.9741096496582031,
|
| 1949 |
+
"step": 21100
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 3.733083865287255,
|
| 1953 |
+
"grad_norm": 15.872313499450684,
|
| 1954 |
+
"learning_rate": 3.7602746962493395e-05,
|
| 1955 |
+
"loss": 0.933817138671875,
|
| 1956 |
+
"step": 21200
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 3.750693374422188,
|
| 1960 |
+
"grad_norm": 9.873473167419434,
|
| 1961 |
+
"learning_rate": 3.749709455890122e-05,
|
| 1962 |
+
"loss": 0.9153755187988282,
|
| 1963 |
+
"step": 21300
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 3.768302883557121,
|
| 1967 |
+
"grad_norm": 3.8525328636169434,
|
| 1968 |
+
"learning_rate": 3.739144215530903e-05,
|
| 1969 |
+
"loss": 0.8332781219482421,
|
| 1970 |
+
"step": 21400
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 3.7859123926920537,
|
| 1974 |
+
"grad_norm": 7.059874534606934,
|
| 1975 |
+
"learning_rate": 3.7285789751716856e-05,
|
| 1976 |
+
"loss": 0.8763603210449219,
|
| 1977 |
+
"step": 21500
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 3.7859123926920537,
|
| 1981 |
+
"eval_loss": 0.3823384940624237,
|
| 1982 |
+
"eval_mIoU": 42.07849087378942,
|
| 1983 |
+
"eval_mean_accuracy": 53.026848381662504,
|
| 1984 |
+
"eval_overall_accuracy": 95.76488247165432,
|
| 1985 |
+
"eval_runtime": 611.8506,
|
| 1986 |
+
"eval_samples_per_second": 1.888,
|
| 1987 |
+
"eval_steps_per_second": 0.945,
|
| 1988 |
+
"step": 21500
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 3.8035219018269864,
|
| 1992 |
+
"grad_norm": 3.2705864906311035,
|
| 1993 |
+
"learning_rate": 3.718013734812467e-05,
|
| 1994 |
+
"loss": 0.957579345703125,
|
| 1995 |
+
"step": 21600
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 3.8211314109619194,
|
| 1999 |
+
"grad_norm": 2.298734664916992,
|
| 2000 |
+
"learning_rate": 3.707448494453249e-05,
|
| 2001 |
+
"loss": 0.939592514038086,
|
| 2002 |
+
"step": 21700
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 3.8387409200968525,
|
| 2006 |
+
"grad_norm": 9.922309875488281,
|
| 2007 |
+
"learning_rate": 3.6968832540940304e-05,
|
| 2008 |
+
"loss": 0.9532173156738282,
|
| 2009 |
+
"step": 21800
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 3.856350429231785,
|
| 2013 |
+
"grad_norm": 65.7906494140625,
|
| 2014 |
+
"learning_rate": 3.686318013734812e-05,
|
| 2015 |
+
"loss": 0.9110845184326172,
|
| 2016 |
+
"step": 21900
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 3.873959938366718,
|
| 2020 |
+
"grad_norm": 5.08626127243042,
|
| 2021 |
+
"learning_rate": 3.675752773375595e-05,
|
| 2022 |
+
"loss": 0.9018997192382813,
|
| 2023 |
+
"step": 22000
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 3.873959938366718,
|
| 2027 |
+
"eval_loss": 0.36784595251083374,
|
| 2028 |
+
"eval_mIoU": 40.651954450635344,
|
| 2029 |
+
"eval_mean_accuracy": 51.59310256241124,
|
| 2030 |
+
"eval_overall_accuracy": 95.59825715564546,
|
| 2031 |
+
"eval_runtime": 620.1319,
|
| 2032 |
+
"eval_samples_per_second": 1.863,
|
| 2033 |
+
"eval_steps_per_second": 0.932,
|
| 2034 |
+
"step": 22000
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 3.891569447501651,
|
| 2038 |
+
"grad_norm": 3.0670690536499023,
|
| 2039 |
+
"learning_rate": 3.665187533016376e-05,
|
| 2040 |
+
"loss": 1.059849166870117,
|
| 2041 |
+
"step": 22100
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 3.909178956636584,
|
| 2045 |
+
"grad_norm": 23.933759689331055,
|
| 2046 |
+
"learning_rate": 3.654622292657158e-05,
|
| 2047 |
+
"loss": 0.8858226013183593,
|
| 2048 |
+
"step": 22200
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 3.9267884657715166,
|
| 2052 |
+
"grad_norm": 2.215423345565796,
|
| 2053 |
+
"learning_rate": 3.6440570522979395e-05,
|
| 2054 |
+
"loss": 0.9452084350585938,
|
| 2055 |
+
"step": 22300
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 3.9443979749064493,
|
| 2059 |
+
"grad_norm": 309.3457336425781,
|
| 2060 |
+
"learning_rate": 3.633491811938722e-05,
|
| 2061 |
+
"loss": 0.9484812927246093,
|
| 2062 |
+
"step": 22400
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 3.9620074840413824,
|
| 2066 |
+
"grad_norm": 2.219196319580078,
|
| 2067 |
+
"learning_rate": 3.622926571579504e-05,
|
| 2068 |
+
"loss": 1.0205068969726563,
|
| 2069 |
+
"step": 22500
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 3.9620074840413824,
|
| 2073 |
+
"eval_loss": 0.3499414622783661,
|
| 2074 |
+
"eval_mIoU": 40.56537017455844,
|
| 2075 |
+
"eval_mean_accuracy": 52.12166106488223,
|
| 2076 |
+
"eval_overall_accuracy": 95.7546244039164,
|
| 2077 |
+
"eval_runtime": 612.0467,
|
| 2078 |
+
"eval_samples_per_second": 1.887,
|
| 2079 |
+
"eval_steps_per_second": 0.944,
|
| 2080 |
+
"step": 22500
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 3.9796169931763155,
|
| 2084 |
+
"grad_norm": 24.519412994384766,
|
| 2085 |
+
"learning_rate": 3.6123613312202856e-05,
|
| 2086 |
+
"loss": 0.9474005889892578,
|
| 2087 |
+
"step": 22600
|
| 2088 |
+
},
|
| 2089 |
+
{
|
| 2090 |
+
"epoch": 3.997226502311248,
|
| 2091 |
+
"grad_norm": 6.02849817276001,
|
| 2092 |
+
"learning_rate": 3.6017960908610674e-05,
|
| 2093 |
+
"loss": 0.9866950225830078,
|
| 2094 |
+
"step": 22700
|
| 2095 |
+
},
|
| 2096 |
+
{
|
| 2097 |
+
"epoch": 4.014791987673344,
|
| 2098 |
+
"grad_norm": 10.79825210571289,
|
| 2099 |
+
"learning_rate": 3.5912308505018486e-05,
|
| 2100 |
+
"loss": 0.7545838928222657,
|
| 2101 |
+
"step": 22800
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 4.032401496808276,
|
| 2105 |
+
"grad_norm": 8.403230667114258,
|
| 2106 |
+
"learning_rate": 3.580665610142631e-05,
|
| 2107 |
+
"loss": 0.8569242095947266,
|
| 2108 |
+
"step": 22900
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 4.050011005943209,
|
| 2112 |
+
"grad_norm": 7.723455905914307,
|
| 2113 |
+
"learning_rate": 3.570100369783412e-05,
|
| 2114 |
+
"loss": 0.6824626922607422,
|
| 2115 |
+
"step": 23000
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 4.050011005943209,
|
| 2119 |
+
"eval_loss": 0.39329513907432556,
|
| 2120 |
+
"eval_mIoU": 41.46868879356731,
|
| 2121 |
+
"eval_mean_accuracy": 51.606976880689324,
|
| 2122 |
+
"eval_overall_accuracy": 95.72947019106381,
|
| 2123 |
+
"eval_runtime": 613.8892,
|
| 2124 |
+
"eval_samples_per_second": 1.881,
|
| 2125 |
+
"eval_steps_per_second": 0.942,
|
| 2126 |
+
"step": 23000
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 4.067620515078143,
|
| 2130 |
+
"grad_norm": 96.14293670654297,
|
| 2131 |
+
"learning_rate": 3.559535129424195e-05,
|
| 2132 |
+
"loss": 0.7759761810302734,
|
| 2133 |
+
"step": 23100
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 4.085230024213075,
|
| 2137 |
+
"grad_norm": 19.32292938232422,
|
| 2138 |
+
"learning_rate": 3.5489698890649765e-05,
|
| 2139 |
+
"loss": 0.7245157623291015,
|
| 2140 |
+
"step": 23200
|
| 2141 |
+
},
|
| 2142 |
+
{
|
| 2143 |
+
"epoch": 4.102839533348008,
|
| 2144 |
+
"grad_norm": 4.930475234985352,
|
| 2145 |
+
"learning_rate": 3.538404648705758e-05,
|
| 2146 |
+
"loss": 0.7503673553466796,
|
| 2147 |
+
"step": 23300
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 4.1204490424829405,
|
| 2151 |
+
"grad_norm": 8.997782707214355,
|
| 2152 |
+
"learning_rate": 3.52783940834654e-05,
|
| 2153 |
+
"loss": 0.7232845306396485,
|
| 2154 |
+
"step": 23400
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 4.138058551617874,
|
| 2158 |
+
"grad_norm": 2.9045639038085938,
|
| 2159 |
+
"learning_rate": 3.517274167987321e-05,
|
| 2160 |
+
"loss": 0.7996186065673828,
|
| 2161 |
+
"step": 23500
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 4.138058551617874,
|
| 2165 |
+
"eval_loss": 0.39504244923591614,
|
| 2166 |
+
"eval_mIoU": 41.049576590313016,
|
| 2167 |
+
"eval_mean_accuracy": 53.10629746092167,
|
| 2168 |
+
"eval_overall_accuracy": 95.63078711043188,
|
| 2169 |
+
"eval_runtime": 622.9242,
|
| 2170 |
+
"eval_samples_per_second": 1.854,
|
| 2171 |
+
"eval_steps_per_second": 0.928,
|
| 2172 |
+
"step": 23500
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 4.155668060752807,
|
| 2176 |
+
"grad_norm": 21.821067810058594,
|
| 2177 |
+
"learning_rate": 3.506708927628104e-05,
|
| 2178 |
+
"loss": 0.7994131469726562,
|
| 2179 |
+
"step": 23600
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 4.173277569887739,
|
| 2183 |
+
"grad_norm": 2.4144773483276367,
|
| 2184 |
+
"learning_rate": 3.496143687268885e-05,
|
| 2185 |
+
"loss": 0.7240392303466797,
|
| 2186 |
+
"step": 23700
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 4.190887079022672,
|
| 2190 |
+
"grad_norm": 5.613311290740967,
|
| 2191 |
+
"learning_rate": 3.4855784469096674e-05,
|
| 2192 |
+
"loss": 0.7988972473144531,
|
| 2193 |
+
"step": 23800
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 4.2084965881576055,
|
| 2197 |
+
"grad_norm": 2.1184654235839844,
|
| 2198 |
+
"learning_rate": 3.475013206550449e-05,
|
| 2199 |
+
"loss": 0.6988441467285156,
|
| 2200 |
+
"step": 23900
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 4.226106097292538,
|
| 2204 |
+
"grad_norm": 3.8888204097747803,
|
| 2205 |
+
"learning_rate": 3.464447966191231e-05,
|
| 2206 |
+
"loss": 0.8608739471435547,
|
| 2207 |
+
"step": 24000
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 4.226106097292538,
|
| 2211 |
+
"eval_loss": 0.4150463938713074,
|
| 2212 |
+
"eval_mIoU": 41.66218151688048,
|
| 2213 |
+
"eval_mean_accuracy": 52.747956293795106,
|
| 2214 |
+
"eval_overall_accuracy": 95.68294574688008,
|
| 2215 |
+
"eval_runtime": 615.4286,
|
| 2216 |
+
"eval_samples_per_second": 1.877,
|
| 2217 |
+
"eval_steps_per_second": 0.939,
|
| 2218 |
+
"step": 24000
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 4.243715606427471,
|
| 2222 |
+
"grad_norm": 1.9162812232971191,
|
| 2223 |
+
"learning_rate": 3.453882725832013e-05,
|
| 2224 |
+
"loss": 0.7657085418701172,
|
| 2225 |
+
"step": 24100
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 4.2613251155624035,
|
| 2229 |
+
"grad_norm": 11.44760513305664,
|
| 2230 |
+
"learning_rate": 3.4433174854727947e-05,
|
| 2231 |
+
"loss": 0.7565657043457031,
|
| 2232 |
+
"step": 24200
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 4.278934624697337,
|
| 2236 |
+
"grad_norm": 11.871203422546387,
|
| 2237 |
+
"learning_rate": 3.4327522451135765e-05,
|
| 2238 |
+
"loss": 0.7799417114257813,
|
| 2239 |
+
"step": 24300
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 4.29654413383227,
|
| 2243 |
+
"grad_norm": 13.05679988861084,
|
| 2244 |
+
"learning_rate": 3.422187004754358e-05,
|
| 2245 |
+
"loss": 0.7182006072998047,
|
| 2246 |
+
"step": 24400
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 4.314153642967202,
|
| 2250 |
+
"grad_norm": 17.702442169189453,
|
| 2251 |
+
"learning_rate": 3.41162176439514e-05,
|
| 2252 |
+
"loss": 0.721866455078125,
|
| 2253 |
+
"step": 24500
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 4.314153642967202,
|
| 2257 |
+
"eval_loss": 0.4177318215370178,
|
| 2258 |
+
"eval_mIoU": 41.28552847096313,
|
| 2259 |
+
"eval_mean_accuracy": 52.55204362802653,
|
| 2260 |
+
"eval_overall_accuracy": 95.7574832800663,
|
| 2261 |
+
"eval_runtime": 615.2573,
|
| 2262 |
+
"eval_samples_per_second": 1.877,
|
| 2263 |
+
"eval_steps_per_second": 0.939,
|
| 2264 |
+
"step": 24500
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 4.331763152102135,
|
| 2268 |
+
"grad_norm": 4.1911492347717285,
|
| 2269 |
+
"learning_rate": 3.401056524035922e-05,
|
| 2270 |
+
"loss": 0.803135986328125,
|
| 2271 |
+
"step": 24600
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 4.349372661237068,
|
| 2275 |
+
"grad_norm": 4.093879222869873,
|
| 2276 |
+
"learning_rate": 3.390491283676704e-05,
|
| 2277 |
+
"loss": 0.8821774291992187,
|
| 2278 |
+
"step": 24700
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 4.366982170372001,
|
| 2282 |
+
"grad_norm": 7.071560859680176,
|
| 2283 |
+
"learning_rate": 3.3799260433174856e-05,
|
| 2284 |
+
"loss": 0.8138728332519531,
|
| 2285 |
+
"step": 24800
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 4.384591679506934,
|
| 2289 |
+
"grad_norm": 6.610219478607178,
|
| 2290 |
+
"learning_rate": 3.3693608029582674e-05,
|
| 2291 |
+
"loss": 0.7457758331298828,
|
| 2292 |
+
"step": 24900
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 4.402201188641866,
|
| 2296 |
+
"grad_norm": 9.474441528320312,
|
| 2297 |
+
"learning_rate": 3.358795562599049e-05,
|
| 2298 |
+
"loss": 0.7298115539550781,
|
| 2299 |
+
"step": 25000
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 4.402201188641866,
|
| 2303 |
+
"eval_loss": 0.38879069685935974,
|
| 2304 |
+
"eval_mIoU": 40.43565637669389,
|
| 2305 |
+
"eval_mean_accuracy": 51.68485905246275,
|
| 2306 |
+
"eval_overall_accuracy": 95.56142798749916,
|
| 2307 |
+
"eval_runtime": 615.8924,
|
| 2308 |
+
"eval_samples_per_second": 1.875,
|
| 2309 |
+
"eval_steps_per_second": 0.938,
|
| 2310 |
+
"step": 25000
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 4.4198106977768,
|
| 2314 |
+
"grad_norm": 16.882156372070312,
|
| 2315 |
+
"learning_rate": 3.348230322239831e-05,
|
| 2316 |
+
"loss": 0.7813544464111328,
|
| 2317 |
+
"step": 25100
|
| 2318 |
+
},
|
| 2319 |
+
{
|
| 2320 |
+
"epoch": 4.437420206911733,
|
| 2321 |
+
"grad_norm": 6.276833534240723,
|
| 2322 |
+
"learning_rate": 3.337665081880613e-05,
|
| 2323 |
+
"loss": 0.8473509216308593,
|
| 2324 |
+
"step": 25200
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 4.455029716046665,
|
| 2328 |
+
"grad_norm": 16.890676498413086,
|
| 2329 |
+
"learning_rate": 3.3270998415213946e-05,
|
| 2330 |
+
"loss": 0.7772435760498047,
|
| 2331 |
+
"step": 25300
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 4.472639225181598,
|
| 2335 |
+
"grad_norm": 35.11220932006836,
|
| 2336 |
+
"learning_rate": 3.3165346011621765e-05,
|
| 2337 |
+
"loss": 0.7341469573974609,
|
| 2338 |
+
"step": 25400
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 4.4902487343165305,
|
| 2342 |
+
"grad_norm": 4.116140365600586,
|
| 2343 |
+
"learning_rate": 3.305969360802958e-05,
|
| 2344 |
+
"loss": 0.8031134033203124,
|
| 2345 |
+
"step": 25500
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 4.4902487343165305,
|
| 2349 |
+
"eval_loss": 0.4111625552177429,
|
| 2350 |
+
"eval_mIoU": 40.938853369665615,
|
| 2351 |
+
"eval_mean_accuracy": 52.64935812387781,
|
| 2352 |
+
"eval_overall_accuracy": 95.67260973381273,
|
| 2353 |
+
"eval_runtime": 617.4648,
|
| 2354 |
+
"eval_samples_per_second": 1.871,
|
| 2355 |
+
"eval_steps_per_second": 0.936,
|
| 2356 |
+
"step": 25500
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 4.507858243451464,
|
| 2360 |
+
"grad_norm": 19.544837951660156,
|
| 2361 |
+
"learning_rate": 3.29540412044374e-05,
|
| 2362 |
+
"loss": 0.6685540008544922,
|
| 2363 |
+
"step": 25600
|
| 2364 |
+
},
|
| 2365 |
+
{
|
| 2366 |
+
"epoch": 4.525467752586397,
|
| 2367 |
+
"grad_norm": 26.763763427734375,
|
| 2368 |
+
"learning_rate": 3.284838880084522e-05,
|
| 2369 |
+
"loss": 0.7029299926757813,
|
| 2370 |
+
"step": 25700
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 4.543077261721329,
|
| 2374 |
+
"grad_norm": 11.457063674926758,
|
| 2375 |
+
"learning_rate": 3.2742736397253044e-05,
|
| 2376 |
+
"loss": 0.7187101745605469,
|
| 2377 |
+
"step": 25800
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 4.560686770856263,
|
| 2381 |
+
"grad_norm": 4.882247447967529,
|
| 2382 |
+
"learning_rate": 3.2637083993660855e-05,
|
| 2383 |
+
"loss": 0.8353469085693359,
|
| 2384 |
+
"step": 25900
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 4.5782962799911955,
|
| 2388 |
+
"grad_norm": 7.698673248291016,
|
| 2389 |
+
"learning_rate": 3.2531431590068674e-05,
|
| 2390 |
+
"loss": 0.7884654998779297,
|
| 2391 |
+
"step": 26000
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 4.5782962799911955,
|
| 2395 |
+
"eval_loss": 0.4020868241786957,
|
| 2396 |
+
"eval_mIoU": 41.061296108771664,
|
| 2397 |
+
"eval_mean_accuracy": 52.12832238084668,
|
| 2398 |
+
"eval_overall_accuracy": 95.63389468502689,
|
| 2399 |
+
"eval_runtime": 615.7111,
|
| 2400 |
+
"eval_samples_per_second": 1.876,
|
| 2401 |
+
"eval_steps_per_second": 0.939,
|
| 2402 |
+
"step": 26000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 4.595905789126128,
|
| 2406 |
+
"grad_norm": 27.729001998901367,
|
| 2407 |
+
"learning_rate": 3.242577918647649e-05,
|
| 2408 |
+
"loss": 0.7050986480712891,
|
| 2409 |
+
"step": 26100
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 4.613515298261061,
|
| 2413 |
+
"grad_norm": 4.676613807678223,
|
| 2414 |
+
"learning_rate": 3.232012678288431e-05,
|
| 2415 |
+
"loss": 0.8271269989013672,
|
| 2416 |
+
"step": 26200
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 4.631124807395993,
|
| 2420 |
+
"grad_norm": 9.141484260559082,
|
| 2421 |
+
"learning_rate": 3.221447437929213e-05,
|
| 2422 |
+
"loss": 0.8618730163574219,
|
| 2423 |
+
"step": 26300
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 4.648734316530927,
|
| 2427 |
+
"grad_norm": 22.31208610534668,
|
| 2428 |
+
"learning_rate": 3.2108821975699946e-05,
|
| 2429 |
+
"loss": 0.9146401977539063,
|
| 2430 |
+
"step": 26400
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 4.66634382566586,
|
| 2434 |
+
"grad_norm": 5.31740665435791,
|
| 2435 |
+
"learning_rate": 3.200316957210777e-05,
|
| 2436 |
+
"loss": 0.7881410217285156,
|
| 2437 |
+
"step": 26500
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 4.66634382566586,
|
| 2441 |
+
"eval_loss": 0.3889371454715729,
|
| 2442 |
+
"eval_mIoU": 39.90579024702969,
|
| 2443 |
+
"eval_mean_accuracy": 49.79534666070949,
|
| 2444 |
+
"eval_overall_accuracy": 95.63201805213829,
|
| 2445 |
+
"eval_runtime": 601.1009,
|
| 2446 |
+
"eval_samples_per_second": 1.921,
|
| 2447 |
+
"eval_steps_per_second": 0.962,
|
| 2448 |
+
"step": 26500
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 4.683953334800792,
|
| 2452 |
+
"grad_norm": 96.29251098632812,
|
| 2453 |
+
"learning_rate": 3.189751716851558e-05,
|
| 2454 |
+
"loss": 0.7128916931152344,
|
| 2455 |
+
"step": 26600
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 4.701562843935726,
|
| 2459 |
+
"grad_norm": 10.139142036437988,
|
| 2460 |
+
"learning_rate": 3.17918647649234e-05,
|
| 2461 |
+
"loss": 0.7628507232666015,
|
| 2462 |
+
"step": 26700
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 4.719172353070658,
|
| 2466 |
+
"grad_norm": 3.24714732170105,
|
| 2467 |
+
"learning_rate": 3.168621236133122e-05,
|
| 2468 |
+
"loss": 0.8892244720458984,
|
| 2469 |
+
"step": 26800
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 4.736781862205591,
|
| 2473 |
+
"grad_norm": 5.197122097015381,
|
| 2474 |
+
"learning_rate": 3.158055995773904e-05,
|
| 2475 |
+
"loss": 0.7311017608642578,
|
| 2476 |
+
"step": 26900
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 4.754391371340524,
|
| 2480 |
+
"grad_norm": 33.49533462524414,
|
| 2481 |
+
"learning_rate": 3.147490755414686e-05,
|
| 2482 |
+
"loss": 0.7185327911376953,
|
| 2483 |
+
"step": 27000
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 4.754391371340524,
|
| 2487 |
+
"eval_loss": 0.423967570066452,
|
| 2488 |
+
"eval_mIoU": 41.53527504353472,
|
| 2489 |
+
"eval_mean_accuracy": 52.98828510157863,
|
| 2490 |
+
"eval_overall_accuracy": 95.73358940355705,
|
| 2491 |
+
"eval_runtime": 593.0491,
|
| 2492 |
+
"eval_samples_per_second": 1.948,
|
| 2493 |
+
"eval_steps_per_second": 0.975,
|
| 2494 |
+
"step": 27000
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 4.772000880475456,
|
| 2498 |
+
"grad_norm": 42.050235748291016,
|
| 2499 |
+
"learning_rate": 3.1369255150554673e-05,
|
| 2500 |
+
"loss": 0.748544921875,
|
| 2501 |
+
"step": 27100
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 4.78961038961039,
|
| 2505 |
+
"grad_norm": 12.394989013671875,
|
| 2506 |
+
"learning_rate": 3.12636027469625e-05,
|
| 2507 |
+
"loss": 0.7695165252685547,
|
| 2508 |
+
"step": 27200
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 4.8072198987453225,
|
| 2512 |
+
"grad_norm": 8.161137580871582,
|
| 2513 |
+
"learning_rate": 3.115795034337031e-05,
|
| 2514 |
+
"loss": 0.7662505340576172,
|
| 2515 |
+
"step": 27300
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 4.824829407880255,
|
| 2519 |
+
"grad_norm": 5.988201141357422,
|
| 2520 |
+
"learning_rate": 3.1052297939778135e-05,
|
| 2521 |
+
"loss": 0.6720768737792969,
|
| 2522 |
+
"step": 27400
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 4.842438917015189,
|
| 2526 |
+
"grad_norm": 15.133417129516602,
|
| 2527 |
+
"learning_rate": 3.0946645536185946e-05,
|
| 2528 |
+
"loss": 0.851641845703125,
|
| 2529 |
+
"step": 27500
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 4.842438917015189,
|
| 2533 |
+
"eval_loss": 0.36105746030807495,
|
| 2534 |
+
"eval_mIoU": 42.808893326574484,
|
| 2535 |
+
"eval_mean_accuracy": 53.29861054528663,
|
| 2536 |
+
"eval_overall_accuracy": 95.81305796965893,
|
| 2537 |
+
"eval_runtime": 592.1747,
|
| 2538 |
+
"eval_samples_per_second": 1.95,
|
| 2539 |
+
"eval_steps_per_second": 0.976,
|
| 2540 |
+
"step": 27500
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 4.860048426150121,
|
| 2544 |
+
"grad_norm": 11.938694953918457,
|
| 2545 |
+
"learning_rate": 3.0840993132593764e-05,
|
| 2546 |
+
"loss": 0.7331785583496093,
|
| 2547 |
+
"step": 27600
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 4.877657935285054,
|
| 2551 |
+
"grad_norm": 6.203324794769287,
|
| 2552 |
+
"learning_rate": 3.073534072900159e-05,
|
| 2553 |
+
"loss": 0.7468854522705078,
|
| 2554 |
+
"step": 27700
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 4.895267444419987,
|
| 2558 |
+
"grad_norm": 18.217580795288086,
|
| 2559 |
+
"learning_rate": 3.06296883254094e-05,
|
| 2560 |
+
"loss": 0.7335443115234375,
|
| 2561 |
+
"step": 27800
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 4.912876953554919,
|
| 2565 |
+
"grad_norm": 21.941572189331055,
|
| 2566 |
+
"learning_rate": 3.0524035921817226e-05,
|
| 2567 |
+
"loss": 0.8398058319091797,
|
| 2568 |
+
"step": 27900
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 4.930486462689853,
|
| 2572 |
+
"grad_norm": 9.89223861694336,
|
| 2573 |
+
"learning_rate": 3.041838351822504e-05,
|
| 2574 |
+
"loss": 0.8591268157958984,
|
| 2575 |
+
"step": 28000
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 4.930486462689853,
|
| 2579 |
+
"eval_loss": 0.37627097964286804,
|
| 2580 |
+
"eval_mIoU": 42.55151434436105,
|
| 2581 |
+
"eval_mean_accuracy": 53.672845497296365,
|
| 2582 |
+
"eval_overall_accuracy": 95.75246142102526,
|
| 2583 |
+
"eval_runtime": 598.4996,
|
| 2584 |
+
"eval_samples_per_second": 1.93,
|
| 2585 |
+
"eval_steps_per_second": 0.966,
|
| 2586 |
+
"step": 28000
|
| 2587 |
+
}
|
| 2588 |
+
],
|
| 2589 |
+
"logging_steps": 100,
|
| 2590 |
+
"max_steps": 56790,
|
| 2591 |
+
"num_input_tokens_seen": 0,
|
| 2592 |
+
"num_train_epochs": 10,
|
| 2593 |
+
"save_steps": 500,
|
| 2594 |
+
"stateful_callbacks": {
|
| 2595 |
+
"TrainerControl": {
|
| 2596 |
+
"args": {
|
| 2597 |
+
"should_epoch_stop": false,
|
| 2598 |
+
"should_evaluate": false,
|
| 2599 |
+
"should_log": false,
|
| 2600 |
+
"should_save": true,
|
| 2601 |
+
"should_training_stop": false
|
| 2602 |
+
},
|
| 2603 |
+
"attributes": {}
|
| 2604 |
+
}
|
| 2605 |
+
},
|
| 2606 |
+
"total_flos": 2.9002130312914797e+19,
|
| 2607 |
+
"train_batch_size": 2,
|
| 2608 |
+
"trial_name": null,
|
| 2609 |
+
"trial_params": null
|
| 2610 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e3caf59f7df69ea0fea6a384f8a7dbd4a73e773dab4a7fab5ec4ee9db372bbf
|
| 3 |
+
size 5137
|