Archive stopped 13k-tokenizer long-decay generator: resumable 37k, best-val 36k, corrected 36k evaluation
Browse files- iclr-debug-tokenizer13k-generator135b-v1/ARCHIVE_MANIFEST.json +300 -0
- iclr-debug-tokenizer13k-generator135b-v1/STOPPED.json +12 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/EVAL_DONE.json +5 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-0.json +0 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-0.samples.json +0 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-1.json +0 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-1.samples.json +0 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-2.json +0 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-2.samples.json +0 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-3.json +0 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-3.samples.json +0 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-4.json +0 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-4.samples.json +0 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/summary.json +35 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/exports/step-36000/ncp.pt +3 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/exports/step-36000/vqvae.pt +3 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/exports/step-37000/ncp.pt +3 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/exports/step-37000/vqvae.pt +3 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/progress.json +14 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/retained-best-val-iter-36000.pt +3 -0
- iclr-debug-tokenizer13k-generator135b-v1/generator/retained-stop-iter-37000.pt +3 -0
- iclr-debug-tokenizer13k-generator135b-v1/git-provenance.json +3 -0
iclr-debug-tokenizer13k-generator135b-v1/ARCHIVE_MANIFEST.json
ADDED
|
@@ -0,0 +1,300 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"files": [
|
| 3 |
+
{
|
| 4 |
+
"path": "STOPPED.json",
|
| 5 |
+
"size": 854,
|
| 6 |
+
"sha256": "1592d1f26395fbf18bcad35392bc580848f7b91509aa802f9807407e3e517e30"
|
| 7 |
+
},
|
| 8 |
+
{
|
| 9 |
+
"path": "manifest.json",
|
| 10 |
+
"size": 28934,
|
| 11 |
+
"sha256": "3c34ea9e1045e4c73623ddaf1aa19784cf5868acdabf3b7b5cf38798ece43084"
|
| 12 |
+
},
|
| 13 |
+
{
|
| 14 |
+
"path": "git-provenance.json",
|
| 15 |
+
"size": 59,
|
| 16 |
+
"sha256": "d1310846ad02e018c04dde9dcd79eb8ac05f0e9a5afd18b359290d0d91bcd142"
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"path": "generator/config.yaml",
|
| 20 |
+
"size": 4621,
|
| 21 |
+
"sha256": "ad8f45e4ae962f65a9375ca8453566de319a990bef121d369f31f405cf73830b"
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"path": "generator/progress.json",
|
| 25 |
+
"size": 409,
|
| 26 |
+
"sha256": "7268856ee298081d79c330ca5b12bd79e8b34e92dc17466ecc14c40209753b34"
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"path": "generator/retained-stop-iter-37000.pt",
|
| 30 |
+
"size": 2565487307,
|
| 31 |
+
"sha256": "772a61e696d295001a4b537f8c58c7dfffe0548d8521b7d32b926dd0578c02b7"
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"path": "generator/retained-best-val-iter-36000.pt",
|
| 35 |
+
"size": 2565487435,
|
| 36 |
+
"sha256": "996781be0095ed7e80e131e9df584bd931060beb64b1db59558fe7131684b50f"
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"path": "generator/exports/step-36000/ncp.pt",
|
| 40 |
+
"size": 427469662,
|
| 41 |
+
"sha256": "7d0b65d49302e6142250f17b399efc13f701d906bcd211e740edfb1846da7ea9"
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"path": "generator/exports/step-36000/vqvae.pt",
|
| 45 |
+
"size": 1300864149,
|
| 46 |
+
"sha256": "df8304968d70d1d0ff8d2e69f02978bdfea798d784e8902659aebb3af58ffa01"
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"path": "generator/exports/step-37000/ncp.pt",
|
| 50 |
+
"size": 427469662,
|
| 51 |
+
"sha256": "8c2fdd6cc055275f24ebb06be69391a95fa359cab0a3314c3cd7190ddce98f94"
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"path": "generator/exports/step-37000/vqvae.pt",
|
| 55 |
+
"size": 1300864085,
|
| 56 |
+
"sha256": "3671c7ba09401520a61e06356054fec917c93fcd16b5aa035ddec1d2a8745d1b"
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"path": "generator/evaluation/step-12875/EVAL_DONE.json",
|
| 60 |
+
"size": 97,
|
| 61 |
+
"sha256": "2fa883e6c3995bb1c7343245b27138124ae26b5764b7f7dd3fe2b1dae48ac384"
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"path": "generator/evaluation/step-12875/random-seed-0.json",
|
| 65 |
+
"size": 135369,
|
| 66 |
+
"sha256": "3366b1f198998f579fa29788827bdc4af40915832611afb20ef0e3a9028c455b"
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"path": "generator/evaluation/step-12875/random-seed-0.samples.json",
|
| 70 |
+
"size": 135746,
|
| 71 |
+
"sha256": "a5d9f04606a7bbc0a218161e28ebb532c09d1ded54fe767cc838d57e088094f9"
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"path": "generator/evaluation/step-12875/random-seed-1.json",
|
| 75 |
+
"size": 137886,
|
| 76 |
+
"sha256": "0768895f698a096a50a11187f933ddf5bfa88e38df81d33fd72a08ec49badc89"
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"path": "generator/evaluation/step-12875/random-seed-1.samples.json",
|
| 80 |
+
"size": 138262,
|
| 81 |
+
"sha256": "9c594e7046c41c2272e23c51faf9c67a4efb0743f241cadd6219ca9d3213050c"
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"path": "generator/evaluation/step-12875/random-seed-2.json",
|
| 85 |
+
"size": 138385,
|
| 86 |
+
"sha256": "bc985a51519e25396cafea442713103957c9233d04de32178f965d0092cf2c7e"
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"path": "generator/evaluation/step-12875/random-seed-2.samples.json",
|
| 90 |
+
"size": 138762,
|
| 91 |
+
"sha256": "45856f0eb9bf78b4ae3c395c3d80210be5c95bda4028ea2e1696a545e74ea576"
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"path": "generator/evaluation/step-12875/random-seed-3.json",
|
| 95 |
+
"size": 137242,
|
| 96 |
+
"sha256": "dfc9bb573b5efa4ddd5906c4c84d15277357bf32cb3819d6b94412eb98adc52c"
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"path": "generator/evaluation/step-12875/random-seed-3.samples.json",
|
| 100 |
+
"size": 137619,
|
| 101 |
+
"sha256": "89b36e3a60cb4b60b1933779747b1a5cc2f95344a73bbbed71048b6c0647c0c3"
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"path": "generator/evaluation/step-12875/random-seed-4.json",
|
| 105 |
+
"size": 139424,
|
| 106 |
+
"sha256": "800dfd857b02c6ed8fd9f54e3d4af51adf619eb67afd827347e74c043b613d3b"
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"path": "generator/evaluation/step-12875/random-seed-4.samples.json",
|
| 110 |
+
"size": 139802,
|
| 111 |
+
"sha256": "2e35084b4b69edcc9c1e3f9c2205e5a1bfc3477387f9870c5cabbe97f96c0ae2"
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"path": "generator/evaluation/step-12875/summary.json",
|
| 115 |
+
"size": 866,
|
| 116 |
+
"sha256": "386634f8709b666fc2523c480d1c0501fab049cb368fb6104e8c35eb62d9ce57"
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"path": "generator/evaluation/step-25750/EVAL_DONE.json",
|
| 120 |
+
"size": 97,
|
| 121 |
+
"sha256": "8c09c167a7014823111474214e1dc7d55933435a3817c760ddb3f76585652652"
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"path": "generator/evaluation/step-25750/random-seed-0.json",
|
| 125 |
+
"size": 129252,
|
| 126 |
+
"sha256": "526faa80be182b1b3892f62efdeb6dd889f569a31117044f5c9e060240fb49bf"
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"path": "generator/evaluation/step-25750/random-seed-0.samples.json",
|
| 130 |
+
"size": 129629,
|
| 131 |
+
"sha256": "23a71090695cc3b1eb23d92293b3aeff655d7415c79044beec6d8afcfa238db6"
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"path": "generator/evaluation/step-25750/random-seed-1.json",
|
| 135 |
+
"size": 132249,
|
| 136 |
+
"sha256": "195428056cb5ffdd13f083549675bd74cb0447f055d2fa65920921fac85bb5c7"
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"path": "generator/evaluation/step-25750/random-seed-1.samples.json",
|
| 140 |
+
"size": 132625,
|
| 141 |
+
"sha256": "00bfd837ceed2a0240e123ee49fe2ab8c0c6c37d5993b523d584ff3c80aa8d4c"
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"path": "generator/evaluation/step-25750/random-seed-2.json",
|
| 145 |
+
"size": 134703,
|
| 146 |
+
"sha256": "9f02c2cf6e0b8dac0272de44ab4131df040acd870fc840a1698b17c856abe23e"
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"path": "generator/evaluation/step-25750/random-seed-2.samples.json",
|
| 150 |
+
"size": 135079,
|
| 151 |
+
"sha256": "bfdfdaf0e8ae4e8fc444c38f0fa1126858c5b90b21623aaeeb0d664c0f2cc170"
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"path": "generator/evaluation/step-25750/random-seed-3.json",
|
| 155 |
+
"size": 131290,
|
| 156 |
+
"sha256": "cac93e106727e4dc0188c14665aaebd188bf0e7cceb2d11f14903e26ed8dd66c"
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"path": "generator/evaluation/step-25750/random-seed-3.samples.json",
|
| 160 |
+
"size": 131667,
|
| 161 |
+
"sha256": "f9f55c874161e51d1885979a87d07238fbce54afc790fbe7605311c075ac7ade"
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"path": "generator/evaluation/step-25750/random-seed-4.json",
|
| 165 |
+
"size": 134335,
|
| 166 |
+
"sha256": "c41179230d64fceb58fdf8e7437f06fdbccfbc5a7622188fb2939f771ec6d48f"
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"path": "generator/evaluation/step-25750/random-seed-4.samples.json",
|
| 170 |
+
"size": 134711,
|
| 171 |
+
"sha256": "869ccedb39be65c5f299ba2479feb1b556efabe5242e5d2b73081ef752489098"
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"path": "generator/evaluation/step-25750/summary.json",
|
| 175 |
+
"size": 870,
|
| 176 |
+
"sha256": "02299bb11f75764d748145b484f08372e7dc85649b88125d2c407ccf7d1c09c6"
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"path": "generator/evaluation/step-36000/EVAL_DONE.json",
|
| 180 |
+
"size": 97,
|
| 181 |
+
"sha256": "c925af2101e63e2d09c83006da71989aac86080d8aae7c81455abe1f435cefc8"
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"path": "generator/evaluation/step-36000/random-seed-0.json",
|
| 185 |
+
"size": 129117,
|
| 186 |
+
"sha256": "8450ca555a2422183cd9c0efb6455a7fea95f4ebbcd899d716587a2e9c36ab69"
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"path": "generator/evaluation/step-36000/random-seed-0.samples.json",
|
| 190 |
+
"size": 129493,
|
| 191 |
+
"sha256": "be9cd00a737e8736f1e0dfa2078dc0642263f1909d3ba4fa8a840eaaa6544449"
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"path": "generator/evaluation/step-36000/random-seed-1.json",
|
| 195 |
+
"size": 132992,
|
| 196 |
+
"sha256": "804f0fa0557dac6b2c525f64b2dce499e1c6126caa47d0c226f6b3c1eba1e1a4"
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"path": "generator/evaluation/step-36000/random-seed-1.samples.json",
|
| 200 |
+
"size": 133367,
|
| 201 |
+
"sha256": "566e3b6c91f894c0da3eabecda6c3264cd3c9ae36efd58585495040ba5ff259a"
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"path": "generator/evaluation/step-36000/random-seed-2.json",
|
| 205 |
+
"size": 133483,
|
| 206 |
+
"sha256": "73359ebde05613eb02c486e827a5ad07407e4c637c98654b0c9862446f6e2367"
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"path": "generator/evaluation/step-36000/random-seed-2.samples.json",
|
| 210 |
+
"size": 133859,
|
| 211 |
+
"sha256": "232d2a1afcbe03344f16e5d7aa0c1bb0e47e18a14e94ec4adb43526aced50e3a"
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"path": "generator/evaluation/step-36000/random-seed-3.json",
|
| 215 |
+
"size": 132056,
|
| 216 |
+
"sha256": "4aca02990c3dd49ba66a56b64fe50e8fb5e4e62ddca145721142228c16539778"
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"path": "generator/evaluation/step-36000/random-seed-3.samples.json",
|
| 220 |
+
"size": 132432,
|
| 221 |
+
"sha256": "5c25abff0c88a2c68078a6b009493fdfdbd975e6682b3f16f8d88a7cda008ce0"
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"path": "generator/evaluation/step-36000/random-seed-4.json",
|
| 225 |
+
"size": 134695,
|
| 226 |
+
"sha256": "22560509f55121fa67f7f99f0a1450e8156737430b56a6ff0284f31d821ba5a9"
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"path": "generator/evaluation/step-36000/random-seed-4.samples.json",
|
| 230 |
+
"size": 135072,
|
| 231 |
+
"sha256": "6a4e32aa4c59d5a9dfa817885944ad1334cbe84d894dd6aa5dbac868ca720c19"
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"path": "generator/evaluation/step-36000/summary.json",
|
| 235 |
+
"size": 867,
|
| 236 |
+
"sha256": "698b609f429df36bb65a48fed34e84fe3ee907bd494e4210f0330e3ed6e04feb"
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"path": "generator/evaluation/step-6000/EVAL_DONE.json",
|
| 240 |
+
"size": 96,
|
| 241 |
+
"sha256": "d7b77cbcd10d11a7df75d9d522c248cac1d28c734ce36b3a81a354f6180c2375"
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"path": "generator/evaluation/step-6000/random-seed-0.json",
|
| 245 |
+
"size": 133433,
|
| 246 |
+
"sha256": "00e12799d101f59d15d5211527f6d474c151b6b4a284851eac46ae0a1f5bc2b7"
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"path": "generator/evaluation/step-6000/random-seed-0.samples.json",
|
| 250 |
+
"size": 133810,
|
| 251 |
+
"sha256": "7571db4096fa53703e280c873d8bb14a09359ca4d7e8b1746fb375f976d70b86"
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"path": "generator/evaluation/step-6000/random-seed-1.json",
|
| 255 |
+
"size": 136450,
|
| 256 |
+
"sha256": "669c39ee740357a9ad2bc0b564026236e3653e06954f76db6deb7a03afabbb9f"
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"path": "generator/evaluation/step-6000/random-seed-1.samples.json",
|
| 260 |
+
"size": 136829,
|
| 261 |
+
"sha256": "be2e3046369902823168484645290ffef0e71d2eb0c6c017598f0e003ffb69b3"
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"path": "generator/evaluation/step-6000/random-seed-2.json",
|
| 265 |
+
"size": 137405,
|
| 266 |
+
"sha256": "e5d38f11cb10b5736acf5e4beac564a32ab1d3ecc7983d4963d70f64b803efb9"
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"path": "generator/evaluation/step-6000/random-seed-2.samples.json",
|
| 270 |
+
"size": 137784,
|
| 271 |
+
"sha256": "5f2caef65a8af6634d18a1924ade9807a727a3d58a341e47575ca33ad8896fb9"
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"path": "generator/evaluation/step-6000/random-seed-3.json",
|
| 275 |
+
"size": 134312,
|
| 276 |
+
"sha256": "4a9284048c4e3622c378c8bf0d2da4ddeac8e6acf51d744028c1ac99d819482a"
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"path": "generator/evaluation/step-6000/random-seed-3.samples.json",
|
| 280 |
+
"size": 134692,
|
| 281 |
+
"sha256": "23c81df6fdf1f8c0466c68480300e5d96ce4179b2db43220ed28f26a2c16e218"
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"path": "generator/evaluation/step-6000/random-seed-4.json",
|
| 285 |
+
"size": 136712,
|
| 286 |
+
"sha256": "87659b9b432e0c6affb596c8e6a9f59141d3826cf006aa436fdda93c6e332cb6"
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"path": "generator/evaluation/step-6000/random-seed-4.samples.json",
|
| 290 |
+
"size": 137091,
|
| 291 |
+
"sha256": "a7af98760756d7c8e5513fdd5bf7ad12544c3792f4831ed3406db94600bca638"
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"path": "generator/evaluation/step-6000/summary.json",
|
| 295 |
+
"size": 865,
|
| 296 |
+
"sha256": "557f8196f07fc093d8f52af5dad3efff278124f2d8b51e789cd8deba3b2140cf"
|
| 297 |
+
}
|
| 298 |
+
],
|
| 299 |
+
"evaluator_sha256": "ab5ffe43a59e8ad6942994250986071327894441312deb8a16e9241dc5d51df9"
|
| 300 |
+
}
|
iclr-debug-tokenizer13k-generator135b-v1/STOPPED.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"stopped_utc": "2026-09-17T22:25:39.669249+00:00",
|
| 3 |
+
"reason": "User requested stopping the long-decay local control and starting a fresh generator with 27B cosine horizon, then minimum LR through 135B.",
|
| 4 |
+
"latest_resumable_step": 37000,
|
| 5 |
+
"best_validation_step": 36000,
|
| 6 |
+
"latest_positions": 38797312000,
|
| 7 |
+
"previous_evaluation_precision": "FP32 weights with TF32 enabled at steps 6000, 12875 and 25750; retained unchanged, not precision-matched to TF32-disabled reports.",
|
| 8 |
+
"corrected_evaluation_step": 36000,
|
| 9 |
+
"source_checkout": "/home/ubuntu/mstok-tokenizer13k-generator",
|
| 10 |
+
"source_commit": "39895726dfbbac2fb3340e08c400a6d59fdb468a",
|
| 11 |
+
"note": "Optimizer updates after saved step 37000 were discarded. Best validation is not necessarily best generation. Controller SIGTERM failure status is intentional, not a spontaneous training error."
|
| 12 |
+
}
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/EVAL_DONE.json
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"step": 36000,
|
| 3 |
+
"reference_dtype": "float32",
|
| 4 |
+
"sampling": "supplied-level0-untruncated"
|
| 5 |
+
}
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-0.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-0.samples.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-1.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-1.samples.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-2.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-2.samples.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-3.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-3.samples.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-4.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/random-seed-4.samples.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
iclr-debug-tokenizer13k-generator135b-v1/generator/evaluation/step-36000/summary.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"protocol": {
|
| 3 |
+
"sampling": "random",
|
| 4 |
+
"temperature": 1.0,
|
| 5 |
+
"top_k": 0,
|
| 6 |
+
"top_p": 1.0,
|
| 7 |
+
"n_provided_levels": 1,
|
| 8 |
+
"n_levels": 16,
|
| 9 |
+
"document_aware_validation_sampling": true
|
| 10 |
+
},
|
| 11 |
+
"reference_model": "gpt2-large",
|
| 12 |
+
"checkpoint": "/home/ubuntu/mstok-results/iclr-debug-tokenizer13k-generator135b-v1/generator/exports/step-36000/ncp.pt",
|
| 13 |
+
"checkpoint_step": 36000,
|
| 14 |
+
"seeds": [
|
| 15 |
+
0,
|
| 16 |
+
1,
|
| 17 |
+
2,
|
| 18 |
+
3,
|
| 19 |
+
4
|
| 20 |
+
],
|
| 21 |
+
"samples_per_seed": 128,
|
| 22 |
+
"total_scored_samples": 640,
|
| 23 |
+
"total_scored_tokens": 161340,
|
| 24 |
+
"per_seed_mean_ppl": {
|
| 25 |
+
"0": 213.1966230529294,
|
| 26 |
+
"1": 212.35814317896538,
|
| 27 |
+
"2": 214.7923185337304,
|
| 28 |
+
"3": 225.30468763837635,
|
| 29 |
+
"4": 215.14674847030918
|
| 30 |
+
},
|
| 31 |
+
"mean_gen_ppl": 216.15970417486213,
|
| 32 |
+
"gen_ppl_se": 2.3426295165858577,
|
| 33 |
+
"mean_of_seed_median_ppl": 202.0521240234375,
|
| 34 |
+
"mean_entropy_nats": 4.53923249177583
|
| 35 |
+
}
|
iclr-debug-tokenizer13k-generator135b-v1/generator/exports/step-36000/ncp.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7d0b65d49302e6142250f17b399efc13f701d906bcd211e740edfb1846da7ea9
|
| 3 |
+
size 427469662
|
iclr-debug-tokenizer13k-generator135b-v1/generator/exports/step-36000/vqvae.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df8304968d70d1d0ff8d2e69f02978bdfea798d784e8902659aebb3af58ffa01
|
| 3 |
+
size 1300864149
|
iclr-debug-tokenizer13k-generator135b-v1/generator/exports/step-37000/ncp.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8c2fdd6cc055275f24ebb06be69391a95fa359cab0a3314c3cd7190ddce98f94
|
| 3 |
+
size 427469662
|
iclr-debug-tokenizer13k-generator135b-v1/generator/exports/step-37000/vqvae.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3671c7ba09401520a61e06356054fec917c93fcd16b5aa035ddec1d2a8745d1b
|
| 3 |
+
size 1300864085
|
iclr-debug-tokenizer13k-generator135b-v1/generator/progress.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"step": 37000,
|
| 3 |
+
"checkpoint": "/home/ubuntu/mstok-results/iclr-debug-tokenizer13k-generator135b-v1/generator/checkpoint-iter-37000.pt",
|
| 4 |
+
"nonpadding_tokens": 38796690262,
|
| 5 |
+
"validation": {
|
| 6 |
+
"loss": 6.8699282979965215,
|
| 7 |
+
"reconstruction_loss": 0.0,
|
| 8 |
+
"vq_loss": 0.0,
|
| 9 |
+
"ncp_loss": 6.8699282979965215,
|
| 10 |
+
"mstok_loss": 0.0,
|
| 11 |
+
"semantic_loss": 0.0,
|
| 12 |
+
"ncp_accuracy": 21.956521739130434
|
| 13 |
+
}
|
| 14 |
+
}
|
iclr-debug-tokenizer13k-generator135b-v1/generator/retained-best-val-iter-36000.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:996781be0095ed7e80e131e9df584bd931060beb64b1db59558fe7131684b50f
|
| 3 |
+
size 2565487435
|
iclr-debug-tokenizer13k-generator135b-v1/generator/retained-stop-iter-37000.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:772a61e696d295001a4b537f8c58c7dfffe0548d8521b7d32b926dd0578c02b7
|
| 3 |
+
size 2565487307
|
iclr-debug-tokenizer13k-generator135b-v1/git-provenance.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"commit": "39895726dfbbac2fb3340e08c400a6d59fdb468a"
|
| 3 |
+
}
|