add A3_2 (completed, step 30000)
Browse files- A3_2/config.json +19 -0
- A3_2/model.safetensors +3 -0
- A3_2/run.json +177 -0
A3_2/config.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"arch": "aspect-d bidirectional masked-diffusion transformer over Mimi tokens",
|
| 3 |
+
"config_id": "A3",
|
| 4 |
+
"seed": 2,
|
| 5 |
+
"width": 384,
|
| 6 |
+
"depth": 12,
|
| 7 |
+
"heads": 6,
|
| 8 |
+
"nonembed_params": 21243264,
|
| 9 |
+
"total_params": 33893376,
|
| 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": 4.571121024641047,
|
| 17 |
+
"gpu_hours": 1.0310891108380424,
|
| 18 |
+
"dtype": "bfloat16"
|
| 19 |
+
}
|
A3_2/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b9908c6efdbe1f21e9127162629d1da59a11a2947b8487ff9b7fa432fa20f4ef
|
| 3 |
+
size 67794992
|
A3_2/run.json
ADDED
|
@@ -0,0 +1,177 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"config": "A3",
|
| 3 |
+
"seed": 2,
|
| 4 |
+
"lr": 0.004,
|
| 5 |
+
"steps": 30000,
|
| 6 |
+
"width": 384,
|
| 7 |
+
"depth": 12,
|
| 8 |
+
"heads": 6,
|
| 9 |
+
"nonembed_params": 21243264,
|
| 10 |
+
"total_params": 33893376,
|
| 11 |
+
"micro_batch": 256,
|
| 12 |
+
"accum": 1,
|
| 13 |
+
"device": "NVIDIA B200",
|
| 14 |
+
"started": 1786000799.8127096,
|
| 15 |
+
"status": "completed",
|
| 16 |
+
"recipe": "coarse",
|
| 17 |
+
"wall_seconds": 3711.9207990169525,
|
| 18 |
+
"gpu_hours": 1.0310891108380424,
|
| 19 |
+
"wall_seconds_this_segment": 3711.920824289322,
|
| 20 |
+
"final_step": 30000,
|
| 21 |
+
"val_hist": [
|
| 22 |
+
{
|
| 23 |
+
"step": 1000,
|
| 24 |
+
"val_loss": 6.071007881967195,
|
| 25 |
+
"wall": 155.91099572181702
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"step": 2000,
|
| 29 |
+
"val_loss": 5.782506469281807,
|
| 30 |
+
"wall": 279.81753873825073
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"step": 3000,
|
| 34 |
+
"val_loss": 5.586054015377423,
|
| 35 |
+
"wall": 410.8446886539459
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"step": 4000,
|
| 39 |
+
"val_loss": 5.469508889151963,
|
| 40 |
+
"wall": 533.9302434921265
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"step": 5000,
|
| 44 |
+
"val_loss": 5.361725283048966,
|
| 45 |
+
"wall": 653.8094389438629
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"step": 6000,
|
| 49 |
+
"val_loss": 5.282088303297436,
|
| 50 |
+
"wall": 775.2870154380798
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"step": 7000,
|
| 54 |
+
"val_loss": 5.23664866070695,
|
| 55 |
+
"wall": 898.6819291114807
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"step": 8000,
|
| 59 |
+
"val_loss": 5.188171573147882,
|
| 60 |
+
"wall": 1019.9028518199921
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"step": 9000,
|
| 64 |
+
"val_loss": 5.15512080356962,
|
| 65 |
+
"wall": 1140.7262256145477
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"step": 10000,
|
| 69 |
+
"val_loss": 5.106514861543506,
|
| 70 |
+
"wall": 1261.468338727951
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"step": 11000,
|
| 74 |
+
"val_loss": 5.081736478892338,
|
| 75 |
+
"wall": 1382.8378834724426
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"step": 12000,
|
| 79 |
+
"val_loss": 5.044231359906003,
|
| 80 |
+
"wall": 1504.450493812561
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"step": 13000,
|
| 84 |
+
"val_loss": 5.016005780646242,
|
| 85 |
+
"wall": 1625.973618030548
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"step": 14000,
|
| 89 |
+
"val_loss": 4.995077998759335,
|
| 90 |
+
"wall": 1750.5855829715729
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"step": 15000,
|
| 94 |
+
"val_loss": 4.959861689782713,
|
| 95 |
+
"wall": 1871.1812407970428
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"step": 16000,
|
| 99 |
+
"val_loss": 4.93695053317545,
|
| 100 |
+
"wall": 1992.6073234081268
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"step": 17000,
|
| 104 |
+
"val_loss": 4.908659515704506,
|
| 105 |
+
"wall": 2113.35170173645
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"step": 18000,
|
| 109 |
+
"val_loss": 4.878984692163998,
|
| 110 |
+
"wall": 2235.705159664154
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"step": 19000,
|
| 114 |
+
"val_loss": 4.845631014647237,
|
| 115 |
+
"wall": 2357.2414894104004
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"step": 20000,
|
| 119 |
+
"val_loss": 4.816521887932155,
|
| 120 |
+
"wall": 2481.574625015259
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"step": 21000,
|
| 124 |
+
"val_loss": 4.78626297218865,
|
| 125 |
+
"wall": 2604.134602546692
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"step": 22000,
|
| 129 |
+
"val_loss": 4.75429121225633,
|
| 130 |
+
"wall": 2730.44309425354
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"step": 23000,
|
| 134 |
+
"val_loss": 4.722450526028513,
|
| 135 |
+
"wall": 2851.8291001319885
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"step": 24000,
|
| 139 |
+
"val_loss": 4.691743498588819,
|
| 140 |
+
"wall": 2975.181205511093
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"step": 25000,
|
| 144 |
+
"val_loss": 4.6615818627168215,
|
| 145 |
+
"wall": 3098.050544023514
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"step": 26000,
|
| 149 |
+
"val_loss": 4.6364590322959005,
|
| 150 |
+
"wall": 3220.632518529892
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"step": 27000,
|
| 154 |
+
"val_loss": 4.608783524892414,
|
| 155 |
+
"wall": 3342.2895863056183
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"step": 28000,
|
| 159 |
+
"val_loss": 4.591598142976129,
|
| 160 |
+
"wall": 3465.662348985672
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"step": 29000,
|
| 164 |
+
"val_loss": 4.5802016774785175,
|
| 165 |
+
"wall": 3588.3210146427155
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"step": 30000,
|
| 169 |
+
"val_loss": 4.571121024641047,
|
| 170 |
+
"wall": 3709.6831986904144
|
| 171 |
+
}
|
| 172 |
+
],
|
| 173 |
+
"tokens_seen": 6950056200,
|
| 174 |
+
"final_train_loss": 4.591052366955446,
|
| 175 |
+
"final_val_loss": 4.571121024641047,
|
| 176 |
+
"finished": 1786004511.8628035
|
| 177 |
+
}
|