Training progress: pq-large step 20000, validation 5.0617;pq-rock-balanced step 8950, validation 5.1269;pq-xl step 1200, validation 6.0299;vocoder-rock step 74550, validation 1.4537

checkpoints/pq-rock-balanced/best/model.json

@@ -352,7 +352,7 @@
       "len": 16384
     }
   ],
-  "step": 7000,
+  "step": 8500,
   "has_optimizer_state": false,
   "training": {
     "batch_size": 8,

checkpoints/pq-rock-balanced/best/tensors.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2c4527600f8b6cd8f78a568e73ba6cc12c98ac1e9bc8a19edf7acb74628fd006
+oid sha256:f4b7dfcae55de3c94e613481f3b3cee2fc10a3f8ce7547c5abf8b1da33a8f407
 size 171433984

checkpoints/pq-rock-balanced/latest/model.json

@@ -352,7 +352,7 @@
       "len": 16384
     }
   ],
-  "step": 7250,
+  "step": 8750,
   "has_optimizer_state": true,
   "training": {
     "batch_size": 8,

checkpoints/pq-rock-balanced/latest/tensors.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:67152fac46ded4d60cc2f88216ee2acd31af5cf1fb1e55a5e752a89c1d58410c
+oid sha256:36ad206ef0515cb25a8ecee6bb76ef992dbd2e29b3b8f4bc0b3780af7bfe1024
 size 514301952

checkpoints/pq-xl/best/model.json

@@ -0,0 +1,571 @@
+{
+  "model": {
+    "width": 768,
+    "heads": 12,
+    "layers": 12,
+    "feed_forward_multiplier": 4,
+    "context_length": 1152,
+    "prompt_bytes": 126,
+    "lyrics_bytes": 256,
+    "codebooks": 16,
+    "codebook_size": 1024
+  },
+  "tensors": [
+    {
+      "name": "token_embedding",
+      "len": 12781824
+    },
+    {
+      "name": "position_embedding",
+      "len": 1081344
+    },
+    {
+      "name": "blocks.0.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.0.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.0.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.0.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.0.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.0.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.0.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.0.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.0.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.0.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.1.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.1.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.1.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.1.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.1.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.1.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.1.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.1.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.1.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.1.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.2.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.2.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.2.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.2.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.2.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.2.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.2.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.2.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.2.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.2.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.3.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.3.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.3.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.3.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.3.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.3.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.3.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.3.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.3.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.3.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.4.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.4.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.4.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.4.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.4.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.4.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.4.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.4.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.4.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.4.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.5.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.5.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.5.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.5.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.5.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.5.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.5.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.5.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.5.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.5.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.6.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.6.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.6.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.6.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.6.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.6.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.6.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.6.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.6.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.6.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.7.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.7.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.7.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.7.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.7.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.7.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.7.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.7.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.7.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.7.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.8.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.8.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.8.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.8.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.8.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.8.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.8.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.8.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.8.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.8.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.9.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.9.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.9.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.9.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.9.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.9.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.9.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.9.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.9.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.9.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.10.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.10.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.10.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.10.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.10.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.10.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.10.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.10.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.10.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.10.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.11.attention_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.11.attention.qkv.weight",
+      "len": 1769472
+    },
+    {
+      "name": "blocks.11.attention.qkv.bias",
+      "len": 2304
+    },
+    {
+      "name": "blocks.11.attention.output.weight",
+      "len": 589824
+    },
+    {
+      "name": "blocks.11.attention.output.bias",
+      "len": 768
+    },
+    {
+      "name": "blocks.11.feed_forward_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "blocks.11.feed_forward.expand.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.11.feed_forward.expand.bias",
+      "len": 3072
+    },
+    {
+      "name": "blocks.11.feed_forward.project.weight",
+      "len": 2359296
+    },
+    {
+      "name": "blocks.11.feed_forward.project.bias",
+      "len": 768
+    },
+    {
+      "name": "final_norm.gain",
+      "len": 768
+    },
+    {
+      "name": "head.weight",
+      "len": 12582912
+    },
+    {
+      "name": "head.bias",
+      "len": 16384
+    }
+  ],
+  "step": 1000,
+  "has_optimizer_state": false,
+  "training": {
+    "batch_size": 12,
+    "checkpoint_every": 250,
+    "extra_sources": [
+      {
+        "cache": "data/cache-fma-rock",
+        "track_filter": {
+          "exclude_any": [],
+          "require_any": []
+        },
+        "weight": 0.5
+      },
+      {
+        "cache": "data/cache-pq-jamendolyrics",
+        "track_filter": {
+          "exclude_any": [],
+          "require_any": []
+        },
+        "weight": 0.5
+      }
+    ],
+    "log_every": 25,
+    "model": {
+      "codebook_size": 1024,
+      "codebooks": 16,
+      "context_length": 1152,
+      "feed_forward_multiplier": 4,
+      "heads": 12,
+      "layers": 12,
+      "lyrics_bytes": 256,
+      "prompt_bytes": 126,
+      "width": 768
+    },
+    "optimizer": {
+      "beta1": 0.8999999761581421,
+      "beta2": 0.949999988079071,
+      "decay_steps": 36000,
+      "epsilon": 9.99999993922529e-9,
+      "final_learning_rate": 0.00002499999936844688,
+      "max_gradient_norm": 1.0,
+      "peak_learning_rate": 0.0002500000118743628,
+      "warmup_steps": 2000,
+      "weight_decay": 0.10000000149011612
+    },
+    "seed": 1994,
+    "sequence_length": 1152,
+    "steps": 36000,
+    "track_filter": {
+      "exclude_any": [],
+      "require_any": []
+    },
+    "validate_every": 1000,
+    "validation_batches": 24
+  }
+}

checkpoints/pq-xl/best/tensors.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89287aeedb6bda1e513871f875f9391d72c26f691d24cd20973277f35d2f07ba
+size 445997056

checkpoints/pq-xl/latest/model.json

@@ -512,7 +512,7 @@
       "len": 16384
     }
   ],
-  "step": 750,
+  "step": 1000,
   "has_optimizer_state": true,
   "training": {
     "batch_size": 12,

checkpoints/pq-xl/latest/tensors.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b85540d719d2318d021d0c3a538136472395fb20f59f519b664629cf498a6a62
+oid sha256:b0f109e7207a0f9b3f98788e6dab79c71542ab0002774f36d8464c5e218579ed
 size 1337991168

metrics/pq-rock-balanced.jsonl

@@ -314,3 +314,66 @@
 {"event":"progress","step":7400,"loss":5.0319138,"loss_per_codebook":[2.9714613,2.8746758,2.6996284,2.7340102,5.401009,5.294485,5.184556,4.9636035,6.124326,6.046406,5.8786526,5.5529037,6.3942485,6.3142996,6.2179947,5.85836],"gradient_norm":0.31362635,"learning_rate":0.00017678246,"positions_per_second":2312.290793396083,"milliseconds_per_step":4871.35961972}
 {"event":"progress","step":7425,"loss":5.2192903,"loss_per_codebook":[3.1002443,2.9599893,2.9157646,2.9038322,5.585554,5.53381,5.0995393,5.291539,6.2458043,6.215374,5.976051,5.969524,6.532914,6.4907846,6.376224,6.3116956],"gradient_norm":0.35912365,"learning_rate":0.0001766216,"positions_per_second":2311.607651858435,"milliseconds_per_step":4872.799236039999}
 {"event":"progress","step":7450,"loss":5.02325,"loss_per_codebook":[2.986234,2.8773623,2.870177,2.3684123,5.466742,5.538192,5.1636934,4.5349674,6.225503,6.2898493,6.026568,5.1524773,6.50828,6.558858,6.373604,5.4310727],"gradient_norm":0.38823736,"learning_rate":0.00017646025,"positions_per_second":2311.9100107922254,"milliseconds_per_step":4872.16195588}
+{"event":"progress","step":7475,"loss":5.0979633,"loss_per_codebook":[3.0078738,2.9028802,2.7560902,2.3311558,5.6744766,5.378656,5.3756986,4.8186326,6.255178,6.203556,6.093606,5.530381,6.4747,6.492667,6.454471,5.8174014],"gradient_norm":0.42426974,"learning_rate":0.00017629846,"positions_per_second":2316.402400579075,"milliseconds_per_step":4862.71297128}
+{"event":"progress","step":7500,"loss":5.190328,"loss_per_codebook":[3.0514927,3.241139,3.0757778,2.9070861,5.6657805,5.741049,5.1828732,4.84086,6.252692,6.397366,5.967569,5.466263,6.5413313,6.574999,6.314868,5.8241105],"gradient_norm":0.40223414,"learning_rate":0.00017613618,"positions_per_second":2312.1858728656744,"milliseconds_per_step":4871.5806684}
+{"event":"validation","step":7500,"loss":5.155277,"loss_per_codebook":[3.1837757,3.0607347,2.838914,2.713981,5.656094,5.6192703,5.245364,4.892548,6.27022,6.3145165,6.0006404,5.492071,6.5164375,6.5465546,6.3332767,5.8000355],"improved":true}
+{"event":"progress","step":7525,"loss":5.2563443,"loss_per_codebook":[3.387417,2.8681684,2.5570974,2.58923,5.7583146,5.655117,5.3776298,5.259326,6.371897,6.377436,6.0651336,5.955763,6.565735,6.578616,6.450734,6.2838893],"gradient_norm":0.34918287,"learning_rate":0.00017597343,"positions_per_second":1785.694387138879,"milliseconds_per_step":6307.91029032}
+{"event":"progress","step":7550,"loss":5.195301,"loss_per_codebook":[3.1580431,2.769677,2.565682,2.8982794,5.447447,5.4424486,5.2453275,5.364726,6.1488934,6.188485,6.110226,6.0364804,6.4439187,6.4844313,6.4451833,6.375561],"gradient_norm":0.3554563,"learning_rate":0.00017581022,"positions_per_second":2311.768305187582,"milliseconds_per_step":4872.4606072}
+{"event":"progress","step":7575,"loss":4.820175,"loss_per_codebook":[2.8198724,2.9069026,2.7864585,2.25425,5.1919923,5.4216447,5.118681,4.1568103,5.92969,6.187717,5.839578,4.6301465,6.2953134,6.4715924,6.2187624,4.8933835],"gradient_norm":0.39412612,"learning_rate":0.00017564654,"positions_per_second":2317.5229744082103,"milliseconds_per_step":4860.36174156}
+{"event":"progress","step":7600,"loss":4.9673066,"loss_per_codebook":[2.8787959,3.018363,2.99203,2.5845723,5.43257,5.498137,5.1763554,4.417332,6.094228,6.20541,5.871155,4.9061275,6.371711,6.491202,6.307294,5.2316194],"gradient_norm":0.38711724,"learning_rate":0.00017548239,"positions_per_second":2312.3137575192604,"milliseconds_per_step":4871.31124112}
+{"event":"progress","step":7625,"loss":5.367294,"loss_per_codebook":[3.310311,3.2636375,2.7231486,2.82113,5.823981,5.7991033,5.438925,5.319658,6.425816,6.4271307,6.1843104,6.1097493,6.653329,6.659343,6.506151,6.4109797],"gradient_norm":0.3712002,"learning_rate":0.00017531778,"positions_per_second":2311.521389422552,"milliseconds_per_step":4872.98108144}
+{"event":"progress","step":7650,"loss":4.993794,"loss_per_codebook":[3.0606945,2.834352,2.474727,2.488259,5.5621815,5.5667853,4.637594,4.747772,6.266799,6.302321,5.773126,5.324835,6.4943957,6.5626287,6.1920094,5.6122284],"gradient_norm":0.35729522,"learning_rate":0.0001751527,"positions_per_second":2311.0017677680003,"milliseconds_per_step":4874.07675628}
+{"event":"progress","step":7675,"loss":5.1993766,"loss_per_codebook":[3.1920958,2.8875103,2.740523,2.7762704,5.631598,5.521749,5.1889076,5.076481,6.1861796,6.276302,6.0523887,5.907687,6.4687963,6.557919,6.419248,6.3063645],"gradient_norm":0.31740308,"learning_rate":0.00017498716,"positions_per_second":2311.921562480461,"milliseconds_per_step":4872.13761176}
+{"event":"progress","step":7700,"loss":5.2387824,"loss_per_codebook":[3.4310575,3.0551553,2.596347,2.6033654,5.7727556,5.6274986,5.2761073,5.08362,6.389846,6.3397093,5.9973044,5.8630867,6.590626,6.5725546,6.4040875,6.217389],"gradient_norm":0.32400435,"learning_rate":0.00017482115,"positions_per_second":2317.319836697275,"milliseconds_per_step":4860.7878039199995}
+{"event":"progress","step":7725,"loss":5.213698,"loss_per_codebook":[3.1086771,2.95488,2.766363,2.8390296,5.545934,5.6288705,5.3083134,5.2428226,6.226774,6.3293257,6.0571127,5.8616567,6.428472,6.5603046,6.402692,6.157938],"gradient_norm":0.30588415,"learning_rate":0.0001746547,"positions_per_second":2312.3127291730825,"milliseconds_per_step":4871.3134075200005}
+{"event":"progress","step":7750,"loss":4.975218,"loss_per_codebook":[3.0267603,2.8006349,2.5331895,2.6901298,5.43259,5.403127,5.1807947,4.3276978,6.1205964,6.2117105,6.010621,5.136963,6.407206,6.4976726,6.3886657,5.435127],"gradient_norm":0.36900783,"learning_rate":0.0001744878,"positions_per_second":2311.4090875941497,"milliseconds_per_step":4873.21783948}
+{"event":"progress","step":7775,"loss":5.2780533,"loss_per_codebook":[3.109216,3.1189065,2.947952,2.7531443,5.607184,5.588338,5.3140326,5.404261,6.3025618,6.2679143,6.0930133,6.0657153,6.556826,6.5592704,6.4404955,6.320016],"gradient_norm":0.35040012,"learning_rate":0.00017432042,"positions_per_second":2264.8758136589545,"milliseconds_per_step":4973.34111304}
+{"event":"progress","step":7800,"loss":5.251499,"loss_per_codebook":[3.23011,3.0757084,2.879388,3.0159802,5.707624,5.6269417,5.4243097,5.153373,6.275044,6.309637,6.0936017,5.703095,6.539281,6.570477,6.42554,5.9938707],"gradient_norm":0.3053611,"learning_rate":0.0001741526,"positions_per_second":2313.1586670660163,"milliseconds_per_step":4869.531935}
+{"event":"progress","step":7825,"loss":4.8848124,"loss_per_codebook":[2.971847,2.8864033,2.4963124,2.3845665,5.3065395,5.355441,5.2418504,4.3304563,6.046594,6.1823187,5.9584303,4.7685637,6.3631124,6.4737496,6.3676243,5.0231876],"gradient_norm":0.37432793,"learning_rate":0.0001739843,"positions_per_second":2311.463635811308,"milliseconds_per_step":4873.10283644}
+{"event":"progress","step":7850,"loss":5.2203975,"loss_per_codebook":[3.1035175,3.1615973,2.8902054,2.8897848,5.5732365,5.6418805,5.4049444,5.1351914,6.240706,6.3169518,6.1029983,5.676894,6.4898653,6.5488186,6.419278,5.9304786],"gradient_norm":0.34388682,"learning_rate":0.00017381556,"positions_per_second":2311.767992691906,"milliseconds_per_step":4872.46126584}
+{"event":"progress","step":7875,"loss":5.0654984,"loss_per_codebook":[2.7294514,2.8414516,2.9719,2.920565,5.3322973,5.1561637,5.214025,5.1328406,5.936996,6.0592203,5.9342365,5.754289,6.409393,6.338952,6.2265244,6.089667],"gradient_norm":0.37508088,"learning_rate":0.00017364637,"positions_per_second":2312.0644727867525,"milliseconds_per_step":4871.83646156}
+{"event":"progress","step":7900,"loss":5.251723,"loss_per_codebook":[3.1093252,3.0309498,2.8341584,2.7972429,5.7228193,5.540129,5.2633805,5.2888393,6.3190336,6.3523316,6.144005,5.8417435,6.5379195,6.602668,6.471631,6.1713815],"gradient_norm":0.3421857,"learning_rate":0.00017347676,"positions_per_second":2316.925519027054,"milliseconds_per_step":4861.61506164}
+{"event":"progress","step":7925,"loss":5.1097226,"loss_per_codebook":[3.139609,3.138833,2.8577034,2.6525037,5.4970613,5.533609,5.074665,4.8932967,6.2377396,6.2946067,5.9028335,5.4746084,6.4987316,6.548618,6.2832174,5.727931],"gradient_norm":0.31603974,"learning_rate":0.00017330665,"positions_per_second":2310.9953022796303,"milliseconds_per_step":4874.09039252}
+{"event":"progress","step":7950,"loss":5.225846,"loss_per_codebook":[3.3075728,3.1939783,2.964423,2.9185078,5.616678,5.6902614,5.192488,5.1112814,6.193437,6.3821197,6.0976434,5.546522,6.5205803,6.5734935,6.4677944,5.836757],"gradient_norm":0.36251283,"learning_rate":0.00017313613,"positions_per_second":2312.217092685618,"milliseconds_per_step":4871.51489176}
+{"event":"progress","step":7975,"loss":5.283552,"loss_per_codebook":[3.3104641,3.2167106,2.8142383,2.7264333,5.814095,5.791622,5.3814607,5.0412607,6.303547,6.4116535,6.092191,5.857568,6.53987,6.6363125,6.4426203,6.1567883],"gradient_norm":0.34486523,"learning_rate":0.00017296514,"positions_per_second":2311.630182020751,"milliseconds_per_step":4872.7517436}
+{"event":"progress","step":8000,"loss":4.6979613,"loss_per_codebook":[2.665917,3.015816,2.2478602,2.232591,5.2967477,5.664827,4.2077355,3.8729556,6.0917664,6.346931,5.6490192,4.3255205,6.428823,6.561629,5.98072,4.5785213],"gradient_norm":0.4094019,"learning_rate":0.00017279372,"positions_per_second":2310.5350590199214,"milliseconds_per_step":4875.0612790000005}
+{"event":"validation","step":8000,"loss":5.143868,"loss_per_codebook":[3.173649,3.049778,2.823699,2.714879,5.6372895,5.608866,5.2290344,4.8851666,6.2599754,6.3051753,5.9851165,5.475719,6.508566,6.535937,6.321558,5.7874913],"improved":true}
+{"event":"progress","step":8025,"loss":5.378659,"loss_per_codebook":[3.557611,3.1382208,2.9614453,2.9533923,5.823284,5.7488317,5.4582133,5.3380213,6.407379,6.4142137,6.1667757,5.9961834,6.593845,6.6567473,6.486138,6.3582425],"gradient_norm":0.32199687,"learning_rate":0.00017262183,"positions_per_second":1782.4446406487164,"milliseconds_per_step":6319.41084908}
+{"event":"progress","step":8050,"loss":5.3103056,"loss_per_codebook":[3.1928184,3.1746187,2.9973922,3.0195127,5.6830153,5.6850677,5.4971895,5.219578,6.3206124,6.3843174,6.193199,5.7891846,6.5699143,6.6096377,6.503403,6.125425],"gradient_norm":0.3635092,"learning_rate":0.00017244951,"positions_per_second":2158.037610045961,"milliseconds_per_step":5219.5568546}
+{"event":"progress","step":8075,"loss":5.1187935,"loss_per_codebook":[3.2390401,2.9542396,2.6767254,2.7384157,5.600076,5.4212027,4.9432387,5.070764,6.1177273,6.2292013,5.9796968,5.64267,6.447875,6.542579,6.3531847,5.944057],"gradient_norm":0.36400095,"learning_rate":0.00017227675,"positions_per_second":2311.463640573591,"milliseconds_per_step":4873.102826400001}
+{"event":"progress","step":8100,"loss":4.9021506,"loss_per_codebook":[2.7218385,3.016746,2.5765584,2.6152887,5.0551295,5.46341,4.75046,4.3442,6.103408,6.2599444,5.8299613,5.088166,6.4270296,6.520118,6.2716084,5.390548],"gradient_norm":0.41123143,"learning_rate":0.00017210352,"positions_per_second":2311.909805631286,"milliseconds_per_step":4872.16238824}
+{"event":"progress","step":8125,"loss":5.352101,"loss_per_codebook":[3.489317,3.0935464,2.4214828,2.5438385,5.9279256,5.7663403,5.3784213,5.410252,6.50556,6.4668574,6.155552,6.152704,6.6633034,6.692257,6.4929256,6.4733286],"gradient_norm":0.35823828,"learning_rate":0.00017192989,"positions_per_second":2317.302125500487,"milliseconds_per_step":4860.82495504}
+{"event":"progress","step":8150,"loss":5.131041,"loss_per_codebook":[2.9257157,3.0013428,2.7437625,2.3485217,5.437207,5.52204,5.3086615,5.0720677,6.171513,6.2668157,6.0639205,5.739296,6.472729,6.5234036,6.4272976,6.0723615],"gradient_norm":0.35153076,"learning_rate":0.00017175582,"positions_per_second":2311.9141615461704,"milliseconds_per_step":4872.15320852}
+{"event":"progress","step":8175,"loss":5.203431,"loss_per_codebook":[3.175979,3.157222,3.0487664,3.0390725,5.4808054,5.4863267,5.236737,5.1822567,6.132919,6.249475,5.98236,5.7289686,6.3988094,6.53666,6.3864937,6.032044],"gradient_norm":0.33983672,"learning_rate":0.00017158128,"positions_per_second":2306.524913164891,"milliseconds_per_step":4883.5371062799995}
+{"event":"progress","step":8200,"loss":4.9837904,"loss_per_codebook":[3.1546009,2.8635414,2.4807878,2.5092013,5.2840223,5.2941923,4.8922343,4.6556535,6.0623355,6.2084174,5.7873507,5.542212,6.339813,6.488006,6.2235246,5.9547606],"gradient_norm":0.4102123,"learning_rate":0.00017140631,"positions_per_second":2325.928306188054,"milliseconds_per_step":4842.797591840001}
+{"event":"progress","step":8225,"loss":5.073121,"loss_per_codebook":[2.9620576,2.8214552,2.672785,2.7337525,5.4088383,5.469757,5.1845107,4.760542,6.1606355,6.2045574,6.0537004,5.568454,6.4128356,6.4462337,6.374216,5.935604],"gradient_norm":0.33515638,"learning_rate":0.00017123093,"positions_per_second":2316.2114510674087,"milliseconds_per_step":4863.11385552}
+{"event":"progress","step":8250,"loss":5.2178354,"loss_per_codebook":[3.244203,3.0708425,2.7378826,2.4984608,5.6908836,5.5622873,5.3672996,5.2075295,6.297031,6.288969,6.082024,5.8103113,6.5438657,6.5124297,6.44955,6.121806],"gradient_norm":0.3416352,"learning_rate":0.00017105512,"positions_per_second":2312.007786014761,"milliseconds_per_step":4871.95591128}
+{"event":"progress","step":8275,"loss":4.984491,"loss_per_codebook":[3.1969247,2.724331,2.7048862,2.5247664,5.5613,5.305235,5.0682826,4.716344,6.2207394,6.1526036,5.7383976,5.32188,6.4518404,6.4480085,6.070109,5.5462093],"gradient_norm":0.33391583,"learning_rate":0.00017087886,"positions_per_second":2266.4059101218163,"milliseconds_per_step":4969.98350988}
+{"event":"progress","step":8300,"loss":5.227788,"loss_per_codebook":[3.2729104,3.068654,2.8942401,3.0610683,5.7802815,5.4563227,5.2122345,5.218219,6.329385,6.161971,5.94546,5.790627,6.5737715,6.4209433,6.375642,6.082877],"gradient_norm":0.3219863,"learning_rate":0.00017070214,"positions_per_second":2313.6759533999198,"milliseconds_per_step":4868.44321628}
+{"event":"progress","step":8325,"loss":5.172492,"loss_per_codebook":[3.0526943,2.8629599,2.5508406,2.799755,5.616426,5.5244703,5.131835,5.2005997,6.206806,6.2459073,5.987121,5.9791946,6.437369,6.5093503,6.350358,6.30418],"gradient_norm":0.35169443,"learning_rate":0.00017052505,"positions_per_second":2314.558238186191,"milliseconds_per_step":4866.58741792}
+{"event":"progress","step":8350,"loss":5.0830593,"loss_per_codebook":[3.2583058,3.122997,2.7940845,2.9510953,5.6076565,5.4891906,5.2795367,4.7369423,6.1729956,6.1835074,6.0348306,5.111532,6.428804,6.4504905,6.386123,5.3208537],"gradient_norm":0.33091766,"learning_rate":0.00017034748,"positions_per_second":2311.720653590351,"milliseconds_per_step":4872.56104344}
+{"event":"progress","step":8375,"loss":5.2307324,"loss_per_codebook":[3.249373,3.1505919,2.8221452,2.7511013,5.5971427,5.639612,5.278551,4.987313,6.236876,6.3701086,6.098257,5.7865224,6.5086927,6.5857506,6.4406257,6.189049],"gradient_norm":0.3351778,"learning_rate":0.00017016951,"positions_per_second":2311.5812472662633,"milliseconds_per_step":4872.85489676}
+{"event":"progress","step":8400,"loss":5.1020756,"loss_per_codebook":[3.0570803,2.9801202,3.0895228,3.0329568,5.307623,5.3063493,5.161483,4.939714,6.0842905,6.112945,5.9846144,5.5061355,6.480635,6.4171367,6.3553677,5.8172417],"gradient_norm":0.35512504,"learning_rate":0.00016999111,"positions_per_second":2311.6721982840354,"milliseconds_per_step":4872.66317792}
+{"event":"progress","step":8425,"loss":5.2503242,"loss_per_codebook":[3.246344,3.0253768,2.7536776,2.6405623,5.7454104,5.6435723,5.267332,5.2556043,6.3606973,6.309933,6.0731597,5.9299254,6.53375,6.555542,6.390905,6.2733884],"gradient_norm":0.29612923,"learning_rate":0.00016981227,"positions_per_second":2317.438026581093,"milliseconds_per_step":4860.5399026}
+{"event":"progress","step":8450,"loss":5.2424955,"loss_per_codebook":[3.2366424,3.1091936,2.7898474,2.847962,5.6551743,5.6912484,5.1638103,5.2685375,6.2042727,6.4181027,5.9499164,5.8903966,6.4617677,6.6251693,6.3536325,6.214264],"gradient_norm":0.32421714,"learning_rate":0.00016963301,"positions_per_second":2311.200794491504,"milliseconds_per_step":4873.65703008}
+{"event":"progress","step":8475,"loss":5.2517304,"loss_per_codebook":[3.1771045,3.0122414,2.6261747,2.7847388,5.571165,5.5346785,5.2986794,5.429642,6.2172685,6.339756,6.0834966,6.0543957,6.4823117,6.5708246,6.458524,6.3866816],"gradient_norm":0.35530102,"learning_rate":0.00016945333,"positions_per_second":2312.208226705364,"milliseconds_per_step":4871.5335712}
+{"event":"progress","step":8500,"loss":5.1619754,"loss_per_codebook":[3.0474224,2.9544735,3.0691688,3.067005,5.5005994,5.43883,5.252275,5.170173,6.1367154,6.086488,5.990184,5.7304163,6.40339,6.3569045,6.354683,6.0328836],"gradient_norm":0.34056416,"learning_rate":0.00016927323,"positions_per_second":2311.9230644097966,"milliseconds_per_step":4872.1344466}
+{"event":"validation","step":8500,"loss":5.12693,"loss_per_codebook":[3.1567442,3.0341728,2.8171997,2.6943836,5.6123776,5.5893636,5.2156987,4.8670907,6.240589,6.288567,5.9628005,5.4523425,6.4959702,6.5244346,6.307993,5.77115],"improved":true}
+{"event":"progress","step":8525,"loss":5.1140704,"loss_per_codebook":[3.038063,2.9998357,2.741349,2.685572,5.6436677,5.5762763,5.262884,4.8680964,6.3486094,6.307255,5.9660234,5.360316,6.544841,6.5270224,6.298232,5.6570797],"gradient_norm":0.3318632,"learning_rate":0.0001690927,"positions_per_second":1785.3995708175967,"milliseconds_per_step":6308.951891839999}
+{"event":"progress","step":8550,"loss":5.155302,"loss_per_codebook":[3.0880196,2.9012377,2.7850797,2.6474574,5.558082,5.5837893,5.10974,4.9401693,6.2180076,6.252717,6.1225486,5.7559466,6.4928975,6.462441,6.4432836,6.1234217],"gradient_norm":0.34544748,"learning_rate":0.00016891179,"positions_per_second":2311.2443650163727,"milliseconds_per_step":4873.56515412}
+{"event":"progress","step":8575,"loss":4.9801927,"loss_per_codebook":[2.6910074,2.6127195,2.6875958,2.421723,5.4357233,5.33682,5.1058493,4.7500353,6.159301,6.0994487,5.891882,5.4827223,6.4631653,6.3922563,6.287637,5.8651967],"gradient_norm":0.40463525,"learning_rate":0.00016873042,"positions_per_second":2312.4914921869517,"milliseconds_per_step":4870.93683936}
+{"event":"progress","step":8600,"loss":5.2283363,"loss_per_codebook":[3.3239255,3.1128962,2.7471159,2.8412616,5.7536373,5.59302,5.099357,5.1541123,6.367789,6.3236985,5.9622426,5.7848535,6.5682745,6.570771,6.335729,6.1146917],"gradient_norm":0.32824126,"learning_rate":0.00016854866,"positions_per_second":2312.091406537477,"milliseconds_per_step":4871.77970912}
+{"event":"progress","step":8625,"loss":5.078504,"loss_per_codebook":[3.0300605,3.0291922,2.748757,2.8617988,5.461202,5.553526,4.962861,4.7615085,6.1080227,6.24631,5.9167867,5.414043,6.3951006,6.4985948,6.338351,5.9299574],"gradient_norm":0.4345837,"learning_rate":0.00016836647,"positions_per_second":2312.1161267203192,"milliseconds_per_step":4871.72762208}
+{"event":"progress","step":8650,"loss":5.0110145,"loss_per_codebook":[2.8319795,2.9573016,2.9445686,2.6907089,5.453581,5.3968105,4.9428124,4.726372,6.12794,6.230143,5.7323494,5.297506,6.4216967,6.494352,6.1953616,5.732748],"gradient_norm":0.44489774,"learning_rate":0.00016818388,"positions_per_second":2317.3257579066,"milliseconds_per_step":4860.775383679999}
+{"event":"progress","step":8675,"loss":5.209111,"loss_per_codebook":[3.2646854,3.1223333,2.8913383,2.7881534,5.713719,5.6729465,5.153691,5.0166044,6.2261014,6.408269,5.985536,5.5946465,6.5250936,6.6189084,6.3827868,5.9809637],"gradient_norm":0.31788802,"learning_rate":0.00016800087,"positions_per_second":2311.810355794573,"milliseconds_per_step":4872.37197972}
+{"event":"progress","step":8700,"loss":5.0728745,"loss_per_codebook":[3.213028,2.964975,2.570748,2.4365556,5.5577025,5.453975,5.0601525,5.00891,6.2419066,6.192351,5.779122,5.6216154,6.517347,6.4189935,6.158464,5.9701457],"gradient_norm":0.28785932,"learning_rate":0.00016781743,"positions_per_second":2311.644723419586,"milliseconds_per_step":4872.72109156}
+{"event":"progress","step":8725,"loss":5.3287306,"loss_per_codebook":[3.4038804,3.184602,2.6865306,2.7687795,5.7829866,5.636188,5.394815,5.3997,6.428764,6.3697677,6.1151724,6.0484643,6.6358,6.6102715,6.4270725,6.366891],"gradient_norm":0.30194628,"learning_rate":0.00016763358,"positions_per_second":2311.5507740004305,"milliseconds_per_step":4872.919135800001}
+{"event":"progress","step":8750,"loss":4.7212186,"loss_per_codebook":[2.8375962,2.824699,2.9368048,2.7950034,5.0780706,4.978436,4.7769947,4.2279663,5.7959743,5.841624,5.5583296,4.7244945,6.1767187,6.1289244,5.8549733,5.002882],"gradient_norm":0.36263552,"learning_rate":0.00016744935,"positions_per_second":2311.858470285276,"milliseconds_per_step":4872.270575719999}
+{"event":"progress","step":8775,"loss":5.1157117,"loss_per_codebook":[3.1441536,3.1238136,2.7206473,2.8329353,5.438421,5.6228347,5.223696,4.820907,6.1474943,6.3546643,6.034436,5.354758,6.4351997,6.566664,6.3848715,5.645886],"gradient_norm":0.31790197,"learning_rate":0.00016726469,"positions_per_second":2268.7598562361445,"milliseconds_per_step":4964.82691592}
+{"event":"progress","step":8800,"loss":4.9259405,"loss_per_codebook":[2.951814,2.9805768,2.5710428,2.7132752,5.220876,5.2676487,4.578347,4.7833686,5.978785,6.053252,5.7314777,5.44696,6.291103,6.332472,6.1384435,5.7756033],"gradient_norm":0.38302052,"learning_rate":0.00016707965,"positions_per_second":2311.8746897099086,"milliseconds_per_step":4872.2363933199995}
+{"event":"progress","step":8825,"loss":4.7567625,"loss_per_codebook":[2.753904,2.4975574,2.4517934,2.3002822,5.259867,5.025328,4.8403764,4.3869886,5.9996643,5.935234,5.678319,5.0093865,6.2576175,6.277596,6.094676,5.3396077],"gradient_norm":0.34253004,"learning_rate":0.0001668942,"positions_per_second":2312.1787786120753,"milliseconds_per_step":4871.59561544}
+{"event":"progress","step":8850,"loss":4.4796796,"loss_per_codebook":[2.8617501,2.745738,2.737137,2.4214265,4.880849,4.758658,4.443638,3.8765764,5.7464447,5.395773,5.140156,4.3961444,6.1308312,5.707179,5.500924,4.931647],"gradient_norm":0.5531837,"learning_rate":0.00016670831,"positions_per_second":2314.0763359117436,"milliseconds_per_step":4867.60087608}
+{"event":"progress","step":8875,"loss":5.055633,"loss_per_codebook":[3.1922922,3.2421799,2.871663,2.8387234,5.622913,5.569899,4.748413,4.7726135,6.1044,6.2456064,5.808663,5.3196883,6.3989344,6.494743,6.080963,5.578435],"gradient_norm":0.329971,"learning_rate":0.00016652208,"positions_per_second":2314.5038331755627,"milliseconds_per_step":4866.70181252}
+{"event":"progress","step":8900,"loss":4.7798405,"loss_per_codebook":[2.9354994,2.945853,2.6200094,2.2262177,5.1199617,5.1372495,4.877326,4.347488,5.864961,5.9659953,5.681286,4.9359293,6.212846,6.3388658,6.013102,5.2548575],"gradient_norm":0.39179298,"learning_rate":0.0001663354,"positions_per_second":2311.844630525995,"milliseconds_per_step":4872.299743359999}
+{"event":"progress","step":8925,"loss":5.2369385,"loss_per_codebook":[3.281925,3.0214217,2.6360126,2.8166742,5.6853805,5.5266356,5.249881,5.316907,6.2375493,6.2807555,6.0055556,5.958543,6.4924245,6.515039,6.3989577,6.367365],"gradient_norm":0.3230463,"learning_rate":0.00016614833,"positions_per_second":2311.53978432247,"milliseconds_per_step":4872.94230296}
+{"event":"progress","step":8950,"loss":4.7958403,"loss_per_codebook":[2.9898362,2.655307,2.3385031,2.3532834,5.2777724,5.2316823,4.8510737,4.4392915,5.9976873,6.075626,5.5873713,4.9670424,6.3322225,6.4239354,5.9628534,5.249951],"gradient_norm":0.3505321,"learning_rate":0.00016596087,"positions_per_second":2311.59566909262,"milliseconds_per_step":4872.82449548}

metrics/pq-xl.jsonl

@@ -30,3 +30,21 @@
 {"event":"progress","step":725,"loss":6.3324227,"loss_per_codebook":[5.91255,5.1631403,4.618419,5.259442,6.76243,6.746292,6.6115446,6.2613893,6.872898,6.8711615,6.7855453,6.3688045,6.9067793,6.899766,6.831441,6.4471583],"gradient_norm":0.49102423,"learning_rate":0.000090625006,"positions_per_second":1000.5077449444558,"milliseconds_per_step":16887.42549508}
 {"event":"progress","step":750,"loss":6.3031154,"loss_per_codebook":[5.7700152,5.0380516,4.614352,4.95387,6.7261567,6.6382833,6.630513,6.377908,6.855444,6.851028,6.7925487,6.4568934,6.890302,6.892346,6.835294,6.5268307],"gradient_norm":0.52858096,"learning_rate":0.00009375001,"positions_per_second":1002.6518252711926,"milliseconds_per_step":16851.31326164}
 {"event":"progress","step":775,"loss":6.2777104,"loss_per_codebook":[5.789319,5.115583,4.5907173,4.7284217,6.75266,6.678874,6.5778456,6.2751145,6.872315,6.851867,6.75444,6.400066,6.8964276,6.9020824,6.8157597,6.441875],"gradient_norm":0.52284384,"learning_rate":0.000096875,"positions_per_second":984.5780505402845,"milliseconds_per_step":17160.6506876}
+{"event":"progress","step":800,"loss":6.0843716,"loss_per_codebook":[5.9935284,5.3105154,4.1679544,3.9632368,6.735316,6.742017,6.0965133,5.917147,6.8827596,6.880929,6.2781487,6.0855618,6.902764,6.905634,6.330003,6.1579194],"gradient_norm":0.5082366,"learning_rate":0.000100000005,"positions_per_second":1000.565308519587,"milliseconds_per_step":16886.45394372}
+{"event":"progress","step":825,"loss":6.1310086,"loss_per_codebook":[5.777315,5.200497,4.768041,4.271866,6.735708,6.631925,6.5638676,5.653049,6.8683047,6.8272667,6.746815,5.681982,6.894468,6.8748035,6.813546,5.7866845],"gradient_norm":0.4796103,"learning_rate":0.00010312501,"positions_per_second":1001.5210730802808,"milliseconds_per_step":16870.33898152}
+{"event":"progress","step":850,"loss":6.2059975,"loss_per_codebook":[5.6468534,5.063973,4.601932,4.7196183,6.7557755,6.6397877,6.406107,6.1615925,6.8863096,6.788805,6.6028585,6.2493134,6.9033303,6.840267,6.686668,6.342763],"gradient_norm":0.45770055,"learning_rate":0.00010625,"positions_per_second":1000.5139532641616,"milliseconds_per_step":16887.3207064}
+{"event":"progress","step":875,"loss":6.122939,"loss_per_codebook":[5.5960526,4.918735,4.2852297,4.07691,6.6851273,6.6192245,6.4806786,5.989908,6.8555093,6.823766,6.6934175,6.1424966,6.896203,6.890742,6.767497,6.245538],"gradient_norm":0.59967315,"learning_rate":0.00010937501,"positions_per_second":1000.93799621172,"milliseconds_per_step":16880.1664678}
+{"event":"progress","step":900,"loss":5.8483887,"loss_per_codebook":[5.3142595,4.714829,4.2955747,3.5633934,6.7074347,6.6544924,6.125119,5.1243725,6.857703,6.840005,6.563278,5.176585,6.894023,6.8837194,6.589289,5.270151],"gradient_norm":0.47653097,"learning_rate":0.000112500005,"positions_per_second":1002.4778572046077,"milliseconds_per_step":16854.23760592}
+{"event":"progress","step":925,"loss":5.556119,"loss_per_codebook":[5.407577,4.6946287,3.9615924,2.955798,6.648915,6.3281083,5.582024,4.6882987,6.8114815,6.4960985,6.1126814,4.7510533,6.873773,6.573946,6.203064,4.808868],"gradient_norm":0.5294774,"learning_rate":0.00011562501,"positions_per_second":1000.8113996988666,"milliseconds_per_step":16882.30170548}
+{"event":"progress","step":950,"loss":6.1077847,"loss_per_codebook":[5.283953,4.3651304,3.9447815,4.00816,6.6702614,6.5460854,6.4364877,6.3537784,6.8391905,6.8121743,6.700068,6.5704985,6.882056,6.8677144,6.775934,6.6682796],"gradient_norm":0.49048793,"learning_rate":0.00011875001,"positions_per_second":999.7219686557864,"milliseconds_per_step":16900.69892404}
+{"event":"progress","step":975,"loss":5.875582,"loss_per_codebook":[5.5403175,4.694674,3.932573,3.8549128,6.6617136,6.488335,6.0801854,5.409788,6.826211,6.7709336,6.37748,5.569441,6.859749,6.854868,6.442857,5.6452813],"gradient_norm":0.5155976,"learning_rate":0.000121875004,"positions_per_second":982.1865461286462,"milliseconds_per_step":17202.434778400002}
+{"event":"progress","step":1000,"loss":6.113945,"loss_per_codebook":[5.1940856,4.4807677,3.983462,3.936675,6.6349173,6.569871,6.474784,6.4134755,6.810375,6.8102646,6.6811795,6.620377,6.8610687,6.873677,6.770919,6.707214],"gradient_norm":0.42177412,"learning_rate":0.000125,"positions_per_second":1000.7508307718036,"milliseconds_per_step":16883.3234812}
+{"event":"validation","step":1000,"loss":6.0299077,"loss_per_codebook":[5.2957187,4.6314797,4.159388,4.1142097,6.6410146,6.566482,6.349327,5.898626,6.825968,6.7983913,6.624165,6.0214295,6.8755665,6.8538227,6.703277,6.119654],"improved":true}
+{"event":"progress","step":1025,"loss":6.0480537,"loss_per_codebook":[4.9984956,4.2048655,3.9226375,4.082245,6.596114,6.4416256,6.3655596,6.3784585,6.813553,6.770695,6.628849,6.52427,6.8499365,6.855052,6.7301345,6.60637],"gradient_norm":0.49988863,"learning_rate":0.00012812502,"positions_per_second":764.0253944553906,"milliseconds_per_step":22114.448188}
+{"event":"progress","step":1050,"loss":5.8620987,"loss_per_codebook":[5.063134,4.441273,3.7110963,3.9282837,6.5833483,6.4844785,5.8428526,5.9005013,6.7886033,6.7408576,6.3568363,6.015984,6.8479586,6.811061,6.1479263,6.129387],"gradient_norm":0.4453429,"learning_rate":0.00013125001,"positions_per_second":1000.8632162965854,"milliseconds_per_step":16881.42767652}
+{"event":"progress","step":1075,"loss":6.136392,"loss_per_codebook":[5.0923576,4.625616,4.304042,4.2582555,6.575621,6.558927,6.503749,6.3128576,6.8116255,6.8029695,6.740486,6.4797616,6.873737,6.8559365,6.811972,6.574366],"gradient_norm":0.46352392,"learning_rate":0.000134375,"positions_per_second":1000.8635119793106,"milliseconds_per_step":16881.42268928}
+{"event":"progress","step":1100,"loss":6.0024304,"loss_per_codebook":[5.200432,3.9611495,3.6754074,3.545987,6.5518694,6.453966,6.3948035,6.3054953,6.808878,6.7853165,6.673608,6.5451803,6.878985,6.855831,6.785236,6.6167502],"gradient_norm":0.5924682,"learning_rate":0.0001375,"positions_per_second":1001.2961118839501,"milliseconds_per_step":16874.12924056}
+{"event":"progress","step":1125,"loss":6.003996,"loss_per_codebook":[4.9151297,4.0057483,3.8306105,3.98763,6.5874686,6.479028,6.396319,6.2440042,6.799429,6.786153,6.6471825,6.410552,6.855876,6.8522854,6.755639,6.510879],"gradient_norm":0.38785988,"learning_rate":0.000140625,"positions_per_second":1000.5956074490002,"milliseconds_per_step":16885.9426068}
+{"event":"progress","step":1150,"loss":6.0108824,"loss_per_codebook":[4.8577356,4.239918,4.137903,3.782533,6.5016565,6.47221,6.3768783,6.2178774,6.766652,6.7783628,6.6698494,6.429378,6.8139534,6.840848,6.759199,6.5291615],"gradient_norm":0.51780033,"learning_rate":0.00014375002,"positions_per_second":1000.7891188012859,"milliseconds_per_step":16882.67756172}
+{"event":"progress","step":1175,"loss":5.6947184,"loss_per_codebook":[4.9271293,4.2883224,3.6686773,3.3976984,6.551331,6.4994845,6.042021,5.0816326,6.7923656,6.781347,6.3703685,5.2144456,6.85701,6.8503714,6.458082,5.3352075],"gradient_norm":0.39717272,"learning_rate":0.00014687501,"positions_per_second":1002.6776496902548,"milliseconds_per_step":16850.8792484}
+{"event":"progress","step":1200,"loss":5.986101,"loss_per_codebook":[4.9113703,4.114026,3.8218641,3.9201877,6.496954,6.445945,6.3269634,6.1978383,6.782292,6.7686043,6.616335,6.4155807,6.8560147,6.853112,6.705696,6.544836],"gradient_norm":0.48715526,"learning_rate":0.00015,"positions_per_second":1000.8632794618436,"milliseconds_per_step":16881.42661112}

plots/pq-rock-balanced.json

@@ -2,9 +2,9 @@
   "run": "pq-rock-balanced",
   "parameters": 42858496,
   "train_tracks": 30054,
-  "latest_step": 7450,
-  "latest_train_loss": 5.0232,
-  "best_validation_loss": 5.1703,
-  "best_validation_step": 7000,
-  "milliseconds_per_step": 4872.2
+  "latest_step": 8950,
+  "latest_train_loss": 4.7958,
+  "best_validation_loss": 5.1269,
+  "best_validation_step": 8500,
+  "milliseconds_per_step": 4872.8
 }

plots/pq-xl.json

@@ -2,9 +2,9 @@
   "run": "pq-xl",
   "parameters": 111499264,
   "train_tracks": 30054,
-  "latest_step": 775,
-  "latest_train_loss": 6.2777,
-  "best_validation_loss": null,
-  "best_validation_step": null,
-  "milliseconds_per_step": 17160.7
+  "latest_step": 1200,
+  "latest_train_loss": 5.9861,
+  "best_validation_loss": 6.0299,
+  "best_validation_step": 1000,
+  "milliseconds_per_step": 16881.4
 }

publish.log

@@ -54,3 +54,13 @@ Uploading... 41/41 files checked, 0/15 uploaded (0.00B transferred), 0 committed
 Uploading... 41/41 files checked, 11/15 uploaded (224MB transferred), 0 committed in 0 commit(s)
 Uploading... 41/41 files checked, 11/15 uploaded (746MB transferred), 0 committed in 0 commit(s)
 Uploading... 41/41 files checked, 11/15 uploaded (1.42GB transferred), 0 committed in 0 commit(s)
+2026-11-04 18:59:45 uploaded
+2026-11-04 21:00:43 uploading: pq-large step 20000, validation 5.0617;pq-rock-balanced step 8950, validation 5.1269;pq-xl step 1200, validation 6.0299;vocoder-rock step 74550, validation 1.4537
+Hint: A new version of huggingface_hub (2.4.0) is available! You are using version 1.28.0.
+To update, run: hf update
+Hint: The `hf-cli` skill is not installed. Run `hf skills add -g --claude` to teach your AI agents how to use the `hf` CLI.
+Found 43 files to upload
+Uploading... 43/43 files checked, 0/16 uploaded (0.00B transferred), 0 committed in 0 commit(s)
+Uploading... 43/43 files checked, 11/16 uploaded (626MB transferred), 0 committed in 0 commit(s)
+Uploading... 43/43 files checked, 11/16 uploaded (1.25GB transferred), 0 committed in 0 commit(s)
+Uploading... 43/43 files checked, 11/16 uploaded (1.90GB transferred), 0 committed in 0 commit(s)