add B4_0 (completed, step 30000)
Browse files- B4_0/config.json +19 -0
- B4_0/model.safetensors +3 -0
- B4_0/run.json +177 -0
B4_0/config.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"arch": "aspect-d bidirectional masked-diffusion transformer over Mimi tokens",
|
| 3 |
+
"config_id": "B4",
|
| 4 |
+
"seed": 0,
|
| 5 |
+
"width": 448,
|
| 6 |
+
"depth": 22,
|
| 7 |
+
"heads": 7,
|
| 8 |
+
"nonembed_params": 53006016,
|
| 9 |
+
"total_params": 67764480,
|
| 10 |
+
"mup_base_width": 256,
|
| 11 |
+
"lr_base": 0.004,
|
| 12 |
+
"steps": 30000,
|
| 13 |
+
"status": "completed",
|
| 14 |
+
"codec": "kyutai/mimi (12.5 Hz, 8 codebooks, 2048+MASK+PAD)",
|
| 15 |
+
"phoneme_vocab_file": "phone_vocab.json",
|
| 16 |
+
"final_val_loss": 4.5215161988725905,
|
| 17 |
+
"gpu_hours": 4.848298084669643,
|
| 18 |
+
"dtype": "bfloat16"
|
| 19 |
+
}
|
B4_0/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:88823dae96a659d891c06a9e49b40ae7c2cbfa20af8ac71fbe7e976943cd3fbb
|
| 3 |
+
size 135542816
|
B4_0/run.json
ADDED
|
@@ -0,0 +1,177 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"config": "B4",
|
| 3 |
+
"seed": 0,
|
| 4 |
+
"lr": 0.004,
|
| 5 |
+
"steps": 30000,
|
| 6 |
+
"width": 448,
|
| 7 |
+
"depth": 22,
|
| 8 |
+
"heads": 7,
|
| 9 |
+
"nonembed_params": 53006016,
|
| 10 |
+
"total_params": 67764480,
|
| 11 |
+
"micro_batch": 128,
|
| 12 |
+
"accum": 2,
|
| 13 |
+
"device": "NVIDIA B200",
|
| 14 |
+
"started": 1785980566.234043,
|
| 15 |
+
"status": "completed",
|
| 16 |
+
"recipe": "coarse",
|
| 17 |
+
"wall_seconds": 17453.873104810715,
|
| 18 |
+
"gpu_hours": 4.848298084669643,
|
| 19 |
+
"wall_seconds_this_segment": 17453.873109340668,
|
| 20 |
+
"final_step": 30000,
|
| 21 |
+
"val_hist": [
|
| 22 |
+
{
|
| 23 |
+
"step": 1000,
|
| 24 |
+
"val_loss": 6.126289662685499,
|
| 25 |
+
"wall": 525.7829875946045
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"step": 2000,
|
| 29 |
+
"val_loss": 5.836386487648779,
|
| 30 |
+
"wall": 1128.671791791916
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"step": 3000,
|
| 34 |
+
"val_loss": 5.662523730802064,
|
| 35 |
+
"wall": 1651.314879655838
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"step": 4000,
|
| 39 |
+
"val_loss": 5.527905530414276,
|
| 40 |
+
"wall": 2178.9276161193848
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"step": 5000,
|
| 44 |
+
"val_loss": 5.425746548495028,
|
| 45 |
+
"wall": 2737.8965752124786
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"step": 6000,
|
| 49 |
+
"val_loss": 5.344057817352991,
|
| 50 |
+
"wall": 3331.4690158367157
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"step": 7000,
|
| 54 |
+
"val_loss": 5.290744788096515,
|
| 55 |
+
"wall": 3969.9588534832
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"step": 8000,
|
| 59 |
+
"val_loss": 5.222889682471529,
|
| 60 |
+
"wall": 4577.679897785187
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"step": 9000,
|
| 64 |
+
"val_loss": 5.1799287011709145,
|
| 65 |
+
"wall": 5131.042224407196
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"step": 10000,
|
| 69 |
+
"val_loss": 5.115018425134087,
|
| 70 |
+
"wall": 5748.653543949127
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"step": 11000,
|
| 74 |
+
"val_loss": 5.076876474496146,
|
| 75 |
+
"wall": 6293.961993217468
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"step": 12000,
|
| 79 |
+
"val_loss": 5.041132399170289,
|
| 80 |
+
"wall": 6862.618046045303
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"step": 13000,
|
| 84 |
+
"val_loss": 4.993429445224102,
|
| 85 |
+
"wall": 7486.324169158936
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"step": 14000,
|
| 89 |
+
"val_loss": 4.968508591346989,
|
| 90 |
+
"wall": 8104.3850610256195
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"step": 15000,
|
| 94 |
+
"val_loss": 4.936022295407587,
|
| 95 |
+
"wall": 8666.740983486176
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"step": 16000,
|
| 99 |
+
"val_loss": 4.911614506815396,
|
| 100 |
+
"wall": 9219.982203960419
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"step": 17000,
|
| 104 |
+
"val_loss": 4.874090303892051,
|
| 105 |
+
"wall": 9798.643358707428
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"step": 18000,
|
| 109 |
+
"val_loss": 4.844846730209142,
|
| 110 |
+
"wall": 10351.55480146408
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"step": 19000,
|
| 114 |
+
"val_loss": 4.816265043506796,
|
| 115 |
+
"wall": 10991.706483364105
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"step": 20000,
|
| 119 |
+
"val_loss": 4.7807910795190365,
|
| 120 |
+
"wall": 11633.218596220016
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"step": 21000,
|
| 124 |
+
"val_loss": 4.746610293470996,
|
| 125 |
+
"wall": 12232.599727392197
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"step": 22000,
|
| 129 |
+
"val_loss": 4.714700325544334,
|
| 130 |
+
"wall": 12823.832135438919
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"step": 23000,
|
| 134 |
+
"val_loss": 4.67787073446017,
|
| 135 |
+
"wall": 13374.5461769104
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"step": 24000,
|
| 139 |
+
"val_loss": 4.6488944689924345,
|
| 140 |
+
"wall": 13954.316857814789
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"step": 25000,
|
| 144 |
+
"val_loss": 4.615192154570253,
|
| 145 |
+
"wall": 14511.689994335175
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"step": 26000,
|
| 149 |
+
"val_loss": 4.588374255877155,
|
| 150 |
+
"wall": 15132.807088375092
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"step": 27000,
|
| 154 |
+
"val_loss": 4.565240860755985,
|
| 155 |
+
"wall": 15766.879757642746
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"step": 28000,
|
| 159 |
+
"val_loss": 4.544897266665595,
|
| 160 |
+
"wall": 16320.771750211716
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"step": 29000,
|
| 164 |
+
"val_loss": 4.530654027487859,
|
| 165 |
+
"wall": 16897.73913669586
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"step": 30000,
|
| 169 |
+
"val_loss": 4.5215161988725905,
|
| 170 |
+
"wall": 17448.301064252853
|
| 171 |
+
}
|
| 172 |
+
],
|
| 173 |
+
"tokens_seen": 6950021264,
|
| 174 |
+
"final_train_loss": 4.56182987691756,
|
| 175 |
+
"final_val_loss": 4.5215161988725905,
|
| 176 |
+
"finished": 1785998020.2288804
|
| 177 |
+
}
|