Upload folder using huggingface_hub
Browse files- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +5 -2099
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 13648432
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:44e8847dbf0ea6293644e45813405a2f151b828ae4f29979af6e82dc49e08850
|
3 |
size 13648432
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 27338810
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:67367306cd1d682d64912fd3a1359d1b7b791e240866f779c437cdcf05081345
|
3 |
size 27338810
|
rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:177435cb64bafc6b524cc1f24325d7a1aeaab10e860cd7ae3ed10d5daf104188
|
3 |
size 14244
|
scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9e82f752be7498a405dfc2fff2a403df3464cd10f7d73510023a768d2c3dc848
|
3 |
size 1064
|
trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
-
"best_metric": 2.
|
3 |
-
"best_model_checkpoint": "/home/sunggeunan/data/ICL/outputs/lora/SKIML-ICL_mrqa_nq_v3/Meta-Llama-3-8B-Instruct-unanswerable-1Q-0U-0C-qa_first/checkpoint-
|
4 |
-
"epoch":
|
5 |
"eval_steps": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -2094,2100 +2094,6 @@
|
|
2094 |
"eval_samples_per_second": 1.004,
|
2095 |
"eval_steps_per_second": 0.252,
|
2096 |
"step": 297
|
2097 |
-
},
|
2098 |
-
{
|
2099 |
-
"epoch": 1.000419639110365,
|
2100 |
-
"grad_norm": 0.5272724628448486,
|
2101 |
-
"learning_rate": 3.504728132387707e-07,
|
2102 |
-
"loss": 2.2234,
|
2103 |
-
"step": 298
|
2104 |
-
},
|
2105 |
-
{
|
2106 |
-
"epoch": 1.0037767519932859,
|
2107 |
-
"grad_norm": 0.4873766303062439,
|
2108 |
-
"learning_rate": 3.498817966903073e-07,
|
2109 |
-
"loss": 2.132,
|
2110 |
-
"step": 299
|
2111 |
-
},
|
2112 |
-
{
|
2113 |
-
"epoch": 1.0071338648762065,
|
2114 |
-
"grad_norm": 0.5091378688812256,
|
2115 |
-
"learning_rate": 3.492907801418439e-07,
|
2116 |
-
"loss": 2.1375,
|
2117 |
-
"step": 300
|
2118 |
-
},
|
2119 |
-
{
|
2120 |
-
"epoch": 1.010490977759127,
|
2121 |
-
"grad_norm": 0.5512698888778687,
|
2122 |
-
"learning_rate": 3.486997635933806e-07,
|
2123 |
-
"loss": 2.1544,
|
2124 |
-
"step": 301
|
2125 |
-
},
|
2126 |
-
{
|
2127 |
-
"epoch": 1.013848090642048,
|
2128 |
-
"grad_norm": 0.5361558794975281,
|
2129 |
-
"learning_rate": 3.481087470449172e-07,
|
2130 |
-
"loss": 2.1154,
|
2131 |
-
"step": 302
|
2132 |
-
},
|
2133 |
-
{
|
2134 |
-
"epoch": 1.0172052035249686,
|
2135 |
-
"grad_norm": 0.5224741101264954,
|
2136 |
-
"learning_rate": 3.475177304964539e-07,
|
2137 |
-
"loss": 2.1329,
|
2138 |
-
"step": 303
|
2139 |
-
},
|
2140 |
-
{
|
2141 |
-
"epoch": 1.0205623164078892,
|
2142 |
-
"grad_norm": 0.5074843168258667,
|
2143 |
-
"learning_rate": 3.469267139479905e-07,
|
2144 |
-
"loss": 2.2165,
|
2145 |
-
"step": 304
|
2146 |
-
},
|
2147 |
-
{
|
2148 |
-
"epoch": 1.0239194292908098,
|
2149 |
-
"grad_norm": 0.5245934724807739,
|
2150 |
-
"learning_rate": 3.4633569739952715e-07,
|
2151 |
-
"loss": 2.1086,
|
2152 |
-
"step": 305
|
2153 |
-
},
|
2154 |
-
{
|
2155 |
-
"epoch": 1.0272765421737307,
|
2156 |
-
"grad_norm": 0.5485753417015076,
|
2157 |
-
"learning_rate": 3.457446808510638e-07,
|
2158 |
-
"loss": 2.1842,
|
2159 |
-
"step": 306
|
2160 |
-
},
|
2161 |
-
{
|
2162 |
-
"epoch": 1.0306336550566513,
|
2163 |
-
"grad_norm": 0.5130212903022766,
|
2164 |
-
"learning_rate": 3.4515366430260045e-07,
|
2165 |
-
"loss": 2.1281,
|
2166 |
-
"step": 307
|
2167 |
-
},
|
2168 |
-
{
|
2169 |
-
"epoch": 1.033990767939572,
|
2170 |
-
"grad_norm": 0.49941399693489075,
|
2171 |
-
"learning_rate": 3.445626477541371e-07,
|
2172 |
-
"loss": 2.1702,
|
2173 |
-
"step": 308
|
2174 |
-
},
|
2175 |
-
{
|
2176 |
-
"epoch": 1.0373478808224927,
|
2177 |
-
"grad_norm": 0.5032735466957092,
|
2178 |
-
"learning_rate": 3.4397163120567375e-07,
|
2179 |
-
"loss": 2.1522,
|
2180 |
-
"step": 309
|
2181 |
-
},
|
2182 |
-
{
|
2183 |
-
"epoch": 1.0407049937054134,
|
2184 |
-
"grad_norm": 0.5410491228103638,
|
2185 |
-
"learning_rate": 3.4338061465721043e-07,
|
2186 |
-
"loss": 2.1986,
|
2187 |
-
"step": 310
|
2188 |
-
},
|
2189 |
-
{
|
2190 |
-
"epoch": 1.044062106588334,
|
2191 |
-
"grad_norm": 0.5421656370162964,
|
2192 |
-
"learning_rate": 3.42789598108747e-07,
|
2193 |
-
"loss": 2.1723,
|
2194 |
-
"step": 311
|
2195 |
-
},
|
2196 |
-
{
|
2197 |
-
"epoch": 1.0474192194712548,
|
2198 |
-
"grad_norm": 0.5225166082382202,
|
2199 |
-
"learning_rate": 3.421985815602837e-07,
|
2200 |
-
"loss": 2.1828,
|
2201 |
-
"step": 312
|
2202 |
-
},
|
2203 |
-
{
|
2204 |
-
"epoch": 1.0507763323541754,
|
2205 |
-
"grad_norm": 0.5110374689102173,
|
2206 |
-
"learning_rate": 3.416075650118203e-07,
|
2207 |
-
"loss": 2.1484,
|
2208 |
-
"step": 313
|
2209 |
-
},
|
2210 |
-
{
|
2211 |
-
"epoch": 1.054133445237096,
|
2212 |
-
"grad_norm": 0.5102293491363525,
|
2213 |
-
"learning_rate": 3.41016548463357e-07,
|
2214 |
-
"loss": 2.1511,
|
2215 |
-
"step": 314
|
2216 |
-
},
|
2217 |
-
{
|
2218 |
-
"epoch": 1.0574905581200167,
|
2219 |
-
"grad_norm": 0.5226995348930359,
|
2220 |
-
"learning_rate": 3.404255319148936e-07,
|
2221 |
-
"loss": 2.173,
|
2222 |
-
"step": 315
|
2223 |
-
},
|
2224 |
-
{
|
2225 |
-
"epoch": 1.0608476710029375,
|
2226 |
-
"grad_norm": 0.5546832084655762,
|
2227 |
-
"learning_rate": 3.398345153664303e-07,
|
2228 |
-
"loss": 2.1441,
|
2229 |
-
"step": 316
|
2230 |
-
},
|
2231 |
-
{
|
2232 |
-
"epoch": 1.0642047838858582,
|
2233 |
-
"grad_norm": 0.5536327958106995,
|
2234 |
-
"learning_rate": 3.392434988179669e-07,
|
2235 |
-
"loss": 2.1534,
|
2236 |
-
"step": 317
|
2237 |
-
},
|
2238 |
-
{
|
2239 |
-
"epoch": 1.0675618967687788,
|
2240 |
-
"grad_norm": 0.5330432653427124,
|
2241 |
-
"learning_rate": 3.386524822695035e-07,
|
2242 |
-
"loss": 2.1448,
|
2243 |
-
"step": 318
|
2244 |
-
},
|
2245 |
-
{
|
2246 |
-
"epoch": 1.0709190096516996,
|
2247 |
-
"grad_norm": 0.49918365478515625,
|
2248 |
-
"learning_rate": 3.3806146572104015e-07,
|
2249 |
-
"loss": 2.1029,
|
2250 |
-
"step": 319
|
2251 |
-
},
|
2252 |
-
{
|
2253 |
-
"epoch": 1.0742761225346202,
|
2254 |
-
"grad_norm": 0.5382150411605835,
|
2255 |
-
"learning_rate": 3.3747044917257683e-07,
|
2256 |
-
"loss": 2.1565,
|
2257 |
-
"step": 320
|
2258 |
-
},
|
2259 |
-
{
|
2260 |
-
"epoch": 1.0776332354175409,
|
2261 |
-
"grad_norm": 0.5065953135490417,
|
2262 |
-
"learning_rate": 3.3687943262411345e-07,
|
2263 |
-
"loss": 2.1437,
|
2264 |
-
"step": 321
|
2265 |
-
},
|
2266 |
-
{
|
2267 |
-
"epoch": 1.0809903483004617,
|
2268 |
-
"grad_norm": 0.5226719379425049,
|
2269 |
-
"learning_rate": 3.3628841607565013e-07,
|
2270 |
-
"loss": 2.2127,
|
2271 |
-
"step": 322
|
2272 |
-
},
|
2273 |
-
{
|
2274 |
-
"epoch": 1.0843474611833823,
|
2275 |
-
"grad_norm": 0.5471811890602112,
|
2276 |
-
"learning_rate": 3.3569739952718675e-07,
|
2277 |
-
"loss": 2.1296,
|
2278 |
-
"step": 323
|
2279 |
-
},
|
2280 |
-
{
|
2281 |
-
"epoch": 1.087704574066303,
|
2282 |
-
"grad_norm": 0.5282257795333862,
|
2283 |
-
"learning_rate": 3.351063829787234e-07,
|
2284 |
-
"loss": 2.1872,
|
2285 |
-
"step": 324
|
2286 |
-
},
|
2287 |
-
{
|
2288 |
-
"epoch": 1.0910616869492236,
|
2289 |
-
"grad_norm": 0.5142350792884827,
|
2290 |
-
"learning_rate": 3.3451536643026e-07,
|
2291 |
-
"loss": 2.1382,
|
2292 |
-
"step": 325
|
2293 |
-
},
|
2294 |
-
{
|
2295 |
-
"epoch": 1.0944187998321444,
|
2296 |
-
"grad_norm": 0.502540647983551,
|
2297 |
-
"learning_rate": 3.339243498817967e-07,
|
2298 |
-
"loss": 2.1134,
|
2299 |
-
"step": 326
|
2300 |
-
},
|
2301 |
-
{
|
2302 |
-
"epoch": 1.097775912715065,
|
2303 |
-
"grad_norm": 0.5727620720863342,
|
2304 |
-
"learning_rate": 3.333333333333333e-07,
|
2305 |
-
"loss": 2.1649,
|
2306 |
-
"step": 327
|
2307 |
-
},
|
2308 |
-
{
|
2309 |
-
"epoch": 1.1011330255979856,
|
2310 |
-
"grad_norm": 0.4994928240776062,
|
2311 |
-
"learning_rate": 3.3274231678487e-07,
|
2312 |
-
"loss": 2.0851,
|
2313 |
-
"step": 328
|
2314 |
-
},
|
2315 |
-
{
|
2316 |
-
"epoch": 1.1044901384809065,
|
2317 |
-
"grad_norm": 0.538866400718689,
|
2318 |
-
"learning_rate": 3.321513002364066e-07,
|
2319 |
-
"loss": 2.1617,
|
2320 |
-
"step": 329
|
2321 |
-
},
|
2322 |
-
{
|
2323 |
-
"epoch": 1.1078472513638271,
|
2324 |
-
"grad_norm": 0.5652467608451843,
|
2325 |
-
"learning_rate": 3.315602836879433e-07,
|
2326 |
-
"loss": 2.1395,
|
2327 |
-
"step": 330
|
2328 |
-
},
|
2329 |
-
{
|
2330 |
-
"epoch": 1.1112043642467477,
|
2331 |
-
"grad_norm": 0.5386678576469421,
|
2332 |
-
"learning_rate": 3.3096926713947985e-07,
|
2333 |
-
"loss": 2.1784,
|
2334 |
-
"step": 331
|
2335 |
-
},
|
2336 |
-
{
|
2337 |
-
"epoch": 1.1145614771296686,
|
2338 |
-
"grad_norm": 0.5307468771934509,
|
2339 |
-
"learning_rate": 3.3037825059101653e-07,
|
2340 |
-
"loss": 2.1523,
|
2341 |
-
"step": 332
|
2342 |
-
},
|
2343 |
-
{
|
2344 |
-
"epoch": 1.1179185900125892,
|
2345 |
-
"grad_norm": 0.5158649682998657,
|
2346 |
-
"learning_rate": 3.2978723404255315e-07,
|
2347 |
-
"loss": 2.1502,
|
2348 |
-
"step": 333
|
2349 |
-
},
|
2350 |
-
{
|
2351 |
-
"epoch": 1.1212757028955098,
|
2352 |
-
"grad_norm": 0.49658921360969543,
|
2353 |
-
"learning_rate": 3.2919621749408983e-07,
|
2354 |
-
"loss": 2.1729,
|
2355 |
-
"step": 334
|
2356 |
-
},
|
2357 |
-
{
|
2358 |
-
"epoch": 1.1246328157784307,
|
2359 |
-
"grad_norm": 0.5137538909912109,
|
2360 |
-
"learning_rate": 3.2860520094562645e-07,
|
2361 |
-
"loss": 2.1631,
|
2362 |
-
"step": 335
|
2363 |
-
},
|
2364 |
-
{
|
2365 |
-
"epoch": 1.1279899286613513,
|
2366 |
-
"grad_norm": 0.5163960456848145,
|
2367 |
-
"learning_rate": 3.2801418439716313e-07,
|
2368 |
-
"loss": 2.1616,
|
2369 |
-
"step": 336
|
2370 |
-
},
|
2371 |
-
{
|
2372 |
-
"epoch": 1.131347041544272,
|
2373 |
-
"grad_norm": 0.528138279914856,
|
2374 |
-
"learning_rate": 3.274231678486997e-07,
|
2375 |
-
"loss": 2.1347,
|
2376 |
-
"step": 337
|
2377 |
-
},
|
2378 |
-
{
|
2379 |
-
"epoch": 1.1347041544271925,
|
2380 |
-
"grad_norm": 0.5045042037963867,
|
2381 |
-
"learning_rate": 3.268321513002364e-07,
|
2382 |
-
"loss": 2.1737,
|
2383 |
-
"step": 338
|
2384 |
-
},
|
2385 |
-
{
|
2386 |
-
"epoch": 1.1380612673101134,
|
2387 |
-
"grad_norm": 0.4872915744781494,
|
2388 |
-
"learning_rate": 3.2624113475177305e-07,
|
2389 |
-
"loss": 2.1644,
|
2390 |
-
"step": 339
|
2391 |
-
},
|
2392 |
-
{
|
2393 |
-
"epoch": 1.141418380193034,
|
2394 |
-
"grad_norm": 0.4652014672756195,
|
2395 |
-
"learning_rate": 3.256501182033097e-07,
|
2396 |
-
"loss": 2.0968,
|
2397 |
-
"step": 340
|
2398 |
-
},
|
2399 |
-
{
|
2400 |
-
"epoch": 1.1447754930759546,
|
2401 |
-
"grad_norm": 0.5411556363105774,
|
2402 |
-
"learning_rate": 3.2505910165484635e-07,
|
2403 |
-
"loss": 2.1138,
|
2404 |
-
"step": 341
|
2405 |
-
},
|
2406 |
-
{
|
2407 |
-
"epoch": 1.1481326059588755,
|
2408 |
-
"grad_norm": 0.544403076171875,
|
2409 |
-
"learning_rate": 3.24468085106383e-07,
|
2410 |
-
"loss": 2.1868,
|
2411 |
-
"step": 342
|
2412 |
-
},
|
2413 |
-
{
|
2414 |
-
"epoch": 1.151489718841796,
|
2415 |
-
"grad_norm": 0.5387755036354065,
|
2416 |
-
"learning_rate": 3.2387706855791966e-07,
|
2417 |
-
"loss": 2.1363,
|
2418 |
-
"step": 343
|
2419 |
-
},
|
2420 |
-
{
|
2421 |
-
"epoch": 1.1548468317247167,
|
2422 |
-
"grad_norm": 0.5694748163223267,
|
2423 |
-
"learning_rate": 3.2328605200945623e-07,
|
2424 |
-
"loss": 2.1721,
|
2425 |
-
"step": 344
|
2426 |
-
},
|
2427 |
-
{
|
2428 |
-
"epoch": 1.1582039446076373,
|
2429 |
-
"grad_norm": 0.5088086128234863,
|
2430 |
-
"learning_rate": 3.226950354609929e-07,
|
2431 |
-
"loss": 2.113,
|
2432 |
-
"step": 345
|
2433 |
-
},
|
2434 |
-
{
|
2435 |
-
"epoch": 1.1615610574905582,
|
2436 |
-
"grad_norm": 0.4980618357658386,
|
2437 |
-
"learning_rate": 3.2210401891252953e-07,
|
2438 |
-
"loss": 2.1537,
|
2439 |
-
"step": 346
|
2440 |
-
},
|
2441 |
-
{
|
2442 |
-
"epoch": 1.1649181703734788,
|
2443 |
-
"grad_norm": 0.5294389724731445,
|
2444 |
-
"learning_rate": 3.215130023640662e-07,
|
2445 |
-
"loss": 2.1432,
|
2446 |
-
"step": 347
|
2447 |
-
},
|
2448 |
-
{
|
2449 |
-
"epoch": 1.1682752832563994,
|
2450 |
-
"grad_norm": 0.5375152826309204,
|
2451 |
-
"learning_rate": 3.2092198581560283e-07,
|
2452 |
-
"loss": 2.0828,
|
2453 |
-
"step": 348
|
2454 |
-
},
|
2455 |
-
{
|
2456 |
-
"epoch": 1.1716323961393202,
|
2457 |
-
"grad_norm": 0.5342891216278076,
|
2458 |
-
"learning_rate": 3.203309692671395e-07,
|
2459 |
-
"loss": 2.1481,
|
2460 |
-
"step": 349
|
2461 |
-
},
|
2462 |
-
{
|
2463 |
-
"epoch": 1.1749895090222409,
|
2464 |
-
"grad_norm": 0.5365825891494751,
|
2465 |
-
"learning_rate": 3.197399527186761e-07,
|
2466 |
-
"loss": 2.1582,
|
2467 |
-
"step": 350
|
2468 |
-
},
|
2469 |
-
{
|
2470 |
-
"epoch": 1.1783466219051615,
|
2471 |
-
"grad_norm": 0.5169309377670288,
|
2472 |
-
"learning_rate": 3.1914893617021275e-07,
|
2473 |
-
"loss": 2.1516,
|
2474 |
-
"step": 351
|
2475 |
-
},
|
2476 |
-
{
|
2477 |
-
"epoch": 1.1817037347880823,
|
2478 |
-
"grad_norm": 0.5114818215370178,
|
2479 |
-
"learning_rate": 3.185579196217494e-07,
|
2480 |
-
"loss": 2.0993,
|
2481 |
-
"step": 352
|
2482 |
-
},
|
2483 |
-
{
|
2484 |
-
"epoch": 1.185060847671003,
|
2485 |
-
"grad_norm": 0.5225831270217896,
|
2486 |
-
"learning_rate": 3.1796690307328605e-07,
|
2487 |
-
"loss": 2.1582,
|
2488 |
-
"step": 353
|
2489 |
-
},
|
2490 |
-
{
|
2491 |
-
"epoch": 1.1884179605539236,
|
2492 |
-
"grad_norm": 0.5083090662956238,
|
2493 |
-
"learning_rate": 3.173758865248227e-07,
|
2494 |
-
"loss": 2.133,
|
2495 |
-
"step": 354
|
2496 |
-
},
|
2497 |
-
{
|
2498 |
-
"epoch": 1.1917750734368444,
|
2499 |
-
"grad_norm": 0.5559499263763428,
|
2500 |
-
"learning_rate": 3.1678486997635936e-07,
|
2501 |
-
"loss": 2.229,
|
2502 |
-
"step": 355
|
2503 |
-
},
|
2504 |
-
{
|
2505 |
-
"epoch": 1.195132186319765,
|
2506 |
-
"grad_norm": 0.5239784121513367,
|
2507 |
-
"learning_rate": 3.1619385342789593e-07,
|
2508 |
-
"loss": 2.128,
|
2509 |
-
"step": 356
|
2510 |
-
},
|
2511 |
-
{
|
2512 |
-
"epoch": 1.1984892992026857,
|
2513 |
-
"grad_norm": 0.5411834120750427,
|
2514 |
-
"learning_rate": 3.156028368794326e-07,
|
2515 |
-
"loss": 2.1608,
|
2516 |
-
"step": 357
|
2517 |
-
},
|
2518 |
-
{
|
2519 |
-
"epoch": 1.2018464120856063,
|
2520 |
-
"grad_norm": 0.5520886778831482,
|
2521 |
-
"learning_rate": 3.1501182033096923e-07,
|
2522 |
-
"loss": 2.1963,
|
2523 |
-
"step": 358
|
2524 |
-
},
|
2525 |
-
{
|
2526 |
-
"epoch": 1.2052035249685271,
|
2527 |
-
"grad_norm": 0.5210337042808533,
|
2528 |
-
"learning_rate": 3.144208037825059e-07,
|
2529 |
-
"loss": 2.1664,
|
2530 |
-
"step": 359
|
2531 |
-
},
|
2532 |
-
{
|
2533 |
-
"epoch": 1.2085606378514477,
|
2534 |
-
"grad_norm": 0.5405945181846619,
|
2535 |
-
"learning_rate": 3.1382978723404253e-07,
|
2536 |
-
"loss": 2.1554,
|
2537 |
-
"step": 360
|
2538 |
-
},
|
2539 |
-
{
|
2540 |
-
"epoch": 1.2119177507343684,
|
2541 |
-
"grad_norm": 0.5365945100784302,
|
2542 |
-
"learning_rate": 3.132387706855792e-07,
|
2543 |
-
"loss": 2.1653,
|
2544 |
-
"step": 361
|
2545 |
-
},
|
2546 |
-
{
|
2547 |
-
"epoch": 1.2152748636172892,
|
2548 |
-
"grad_norm": 0.5147272348403931,
|
2549 |
-
"learning_rate": 3.1264775413711583e-07,
|
2550 |
-
"loss": 2.1194,
|
2551 |
-
"step": 362
|
2552 |
-
},
|
2553 |
-
{
|
2554 |
-
"epoch": 1.2186319765002098,
|
2555 |
-
"grad_norm": 0.530087411403656,
|
2556 |
-
"learning_rate": 3.1205673758865245e-07,
|
2557 |
-
"loss": 2.2029,
|
2558 |
-
"step": 363
|
2559 |
-
},
|
2560 |
-
{
|
2561 |
-
"epoch": 1.2219890893831304,
|
2562 |
-
"grad_norm": 0.5508552193641663,
|
2563 |
-
"learning_rate": 3.114657210401891e-07,
|
2564 |
-
"loss": 2.1886,
|
2565 |
-
"step": 364
|
2566 |
-
},
|
2567 |
-
{
|
2568 |
-
"epoch": 1.2253462022660513,
|
2569 |
-
"grad_norm": 0.5228819847106934,
|
2570 |
-
"learning_rate": 3.1087470449172575e-07,
|
2571 |
-
"loss": 2.1737,
|
2572 |
-
"step": 365
|
2573 |
-
},
|
2574 |
-
{
|
2575 |
-
"epoch": 1.228703315148972,
|
2576 |
-
"grad_norm": 0.549346923828125,
|
2577 |
-
"learning_rate": 3.102836879432624e-07,
|
2578 |
-
"loss": 2.1344,
|
2579 |
-
"step": 366
|
2580 |
-
},
|
2581 |
-
{
|
2582 |
-
"epoch": 1.2320604280318925,
|
2583 |
-
"grad_norm": 0.5160160660743713,
|
2584 |
-
"learning_rate": 3.0969267139479906e-07,
|
2585 |
-
"loss": 2.1429,
|
2586 |
-
"step": 367
|
2587 |
-
},
|
2588 |
-
{
|
2589 |
-
"epoch": 1.2354175409148134,
|
2590 |
-
"grad_norm": 0.536289393901825,
|
2591 |
-
"learning_rate": 3.091016548463357e-07,
|
2592 |
-
"loss": 2.154,
|
2593 |
-
"step": 368
|
2594 |
-
},
|
2595 |
-
{
|
2596 |
-
"epoch": 1.238774653797734,
|
2597 |
-
"grad_norm": 0.5272290110588074,
|
2598 |
-
"learning_rate": 3.085106382978723e-07,
|
2599 |
-
"loss": 2.1581,
|
2600 |
-
"step": 369
|
2601 |
-
},
|
2602 |
-
{
|
2603 |
-
"epoch": 1.2421317666806546,
|
2604 |
-
"grad_norm": 0.5044741034507751,
|
2605 |
-
"learning_rate": 3.0791962174940893e-07,
|
2606 |
-
"loss": 2.1434,
|
2607 |
-
"step": 370
|
2608 |
-
},
|
2609 |
-
{
|
2610 |
-
"epoch": 1.2454888795635752,
|
2611 |
-
"grad_norm": 0.5435929298400879,
|
2612 |
-
"learning_rate": 3.073286052009456e-07,
|
2613 |
-
"loss": 2.1461,
|
2614 |
-
"step": 371
|
2615 |
-
},
|
2616 |
-
{
|
2617 |
-
"epoch": 1.248845992446496,
|
2618 |
-
"grad_norm": 0.5109124779701233,
|
2619 |
-
"learning_rate": 3.067375886524823e-07,
|
2620 |
-
"loss": 2.1153,
|
2621 |
-
"step": 372
|
2622 |
-
},
|
2623 |
-
{
|
2624 |
-
"epoch": 1.2522031053294167,
|
2625 |
-
"grad_norm": 0.5592184066772461,
|
2626 |
-
"learning_rate": 3.061465721040189e-07,
|
2627 |
-
"loss": 2.169,
|
2628 |
-
"step": 373
|
2629 |
-
},
|
2630 |
-
{
|
2631 |
-
"epoch": 1.2555602182123373,
|
2632 |
-
"grad_norm": 0.5357188582420349,
|
2633 |
-
"learning_rate": 3.055555555555556e-07,
|
2634 |
-
"loss": 2.1458,
|
2635 |
-
"step": 374
|
2636 |
-
},
|
2637 |
-
{
|
2638 |
-
"epoch": 1.2589173310952582,
|
2639 |
-
"grad_norm": 0.5415797233581543,
|
2640 |
-
"learning_rate": 3.049645390070922e-07,
|
2641 |
-
"loss": 2.2049,
|
2642 |
-
"step": 375
|
2643 |
-
},
|
2644 |
-
{
|
2645 |
-
"epoch": 1.2622744439781788,
|
2646 |
-
"grad_norm": 0.5439038872718811,
|
2647 |
-
"learning_rate": 3.0437352245862883e-07,
|
2648 |
-
"loss": 2.1257,
|
2649 |
-
"step": 376
|
2650 |
-
},
|
2651 |
-
{
|
2652 |
-
"epoch": 1.2656315568610994,
|
2653 |
-
"grad_norm": 0.5348344445228577,
|
2654 |
-
"learning_rate": 3.0378250591016545e-07,
|
2655 |
-
"loss": 2.1166,
|
2656 |
-
"step": 377
|
2657 |
-
},
|
2658 |
-
{
|
2659 |
-
"epoch": 1.26898866974402,
|
2660 |
-
"grad_norm": 0.5547958016395569,
|
2661 |
-
"learning_rate": 3.0319148936170213e-07,
|
2662 |
-
"loss": 2.1121,
|
2663 |
-
"step": 378
|
2664 |
-
},
|
2665 |
-
{
|
2666 |
-
"epoch": 1.2723457826269409,
|
2667 |
-
"grad_norm": 0.5206843018531799,
|
2668 |
-
"learning_rate": 3.0260047281323876e-07,
|
2669 |
-
"loss": 2.1154,
|
2670 |
-
"step": 379
|
2671 |
-
},
|
2672 |
-
{
|
2673 |
-
"epoch": 1.2757028955098615,
|
2674 |
-
"grad_norm": 0.5301013588905334,
|
2675 |
-
"learning_rate": 3.0200945626477543e-07,
|
2676 |
-
"loss": 2.175,
|
2677 |
-
"step": 380
|
2678 |
-
},
|
2679 |
-
{
|
2680 |
-
"epoch": 1.2790600083927823,
|
2681 |
-
"grad_norm": 0.5277062654495239,
|
2682 |
-
"learning_rate": 3.0141843971631206e-07,
|
2683 |
-
"loss": 2.1044,
|
2684 |
-
"step": 381
|
2685 |
-
},
|
2686 |
-
{
|
2687 |
-
"epoch": 1.282417121275703,
|
2688 |
-
"grad_norm": 0.585447371006012,
|
2689 |
-
"learning_rate": 3.008274231678487e-07,
|
2690 |
-
"loss": 2.148,
|
2691 |
-
"step": 382
|
2692 |
-
},
|
2693 |
-
{
|
2694 |
-
"epoch": 1.2857742341586236,
|
2695 |
-
"grad_norm": 0.5090722441673279,
|
2696 |
-
"learning_rate": 3.002364066193853e-07,
|
2697 |
-
"loss": 2.1372,
|
2698 |
-
"step": 383
|
2699 |
-
},
|
2700 |
-
{
|
2701 |
-
"epoch": 1.2891313470415442,
|
2702 |
-
"grad_norm": 0.5260941982269287,
|
2703 |
-
"learning_rate": 2.99645390070922e-07,
|
2704 |
-
"loss": 2.1102,
|
2705 |
-
"step": 384
|
2706 |
-
},
|
2707 |
-
{
|
2708 |
-
"epoch": 1.292488459924465,
|
2709 |
-
"grad_norm": 0.5579008460044861,
|
2710 |
-
"learning_rate": 2.990543735224586e-07,
|
2711 |
-
"loss": 2.1315,
|
2712 |
-
"step": 385
|
2713 |
-
},
|
2714 |
-
{
|
2715 |
-
"epoch": 1.2958455728073857,
|
2716 |
-
"grad_norm": 0.5250607132911682,
|
2717 |
-
"learning_rate": 2.984633569739953e-07,
|
2718 |
-
"loss": 2.1774,
|
2719 |
-
"step": 386
|
2720 |
-
},
|
2721 |
-
{
|
2722 |
-
"epoch": 1.2992026856903063,
|
2723 |
-
"grad_norm": 0.5178482532501221,
|
2724 |
-
"learning_rate": 2.978723404255319e-07,
|
2725 |
-
"loss": 2.0698,
|
2726 |
-
"step": 387
|
2727 |
-
},
|
2728 |
-
{
|
2729 |
-
"epoch": 1.3025597985732271,
|
2730 |
-
"grad_norm": 0.523533821105957,
|
2731 |
-
"learning_rate": 2.9728132387706853e-07,
|
2732 |
-
"loss": 2.1363,
|
2733 |
-
"step": 388
|
2734 |
-
},
|
2735 |
-
{
|
2736 |
-
"epoch": 1.3059169114561477,
|
2737 |
-
"grad_norm": 0.536793053150177,
|
2738 |
-
"learning_rate": 2.9669030732860516e-07,
|
2739 |
-
"loss": 2.1636,
|
2740 |
-
"step": 389
|
2741 |
-
},
|
2742 |
-
{
|
2743 |
-
"epoch": 1.3092740243390684,
|
2744 |
-
"grad_norm": 0.5273007154464722,
|
2745 |
-
"learning_rate": 2.9609929078014183e-07,
|
2746 |
-
"loss": 2.1524,
|
2747 |
-
"step": 390
|
2748 |
-
},
|
2749 |
-
{
|
2750 |
-
"epoch": 1.312631137221989,
|
2751 |
-
"grad_norm": 0.5152904391288757,
|
2752 |
-
"learning_rate": 2.9550827423167846e-07,
|
2753 |
-
"loss": 2.1616,
|
2754 |
-
"step": 391
|
2755 |
-
},
|
2756 |
-
{
|
2757 |
-
"epoch": 1.3159882501049098,
|
2758 |
-
"grad_norm": 0.5256254076957703,
|
2759 |
-
"learning_rate": 2.9491725768321513e-07,
|
2760 |
-
"loss": 2.1526,
|
2761 |
-
"step": 392
|
2762 |
-
},
|
2763 |
-
{
|
2764 |
-
"epoch": 1.3193453629878305,
|
2765 |
-
"grad_norm": 0.5300120711326599,
|
2766 |
-
"learning_rate": 2.9432624113475176e-07,
|
2767 |
-
"loss": 2.1807,
|
2768 |
-
"step": 393
|
2769 |
-
},
|
2770 |
-
{
|
2771 |
-
"epoch": 1.322702475870751,
|
2772 |
-
"grad_norm": 0.5391471982002258,
|
2773 |
-
"learning_rate": 2.9373522458628843e-07,
|
2774 |
-
"loss": 2.1394,
|
2775 |
-
"step": 394
|
2776 |
-
},
|
2777 |
-
{
|
2778 |
-
"epoch": 1.326059588753672,
|
2779 |
-
"grad_norm": 0.5455812811851501,
|
2780 |
-
"learning_rate": 2.93144208037825e-07,
|
2781 |
-
"loss": 2.1385,
|
2782 |
-
"step": 395
|
2783 |
-
},
|
2784 |
-
{
|
2785 |
-
"epoch": 1.3294167016365925,
|
2786 |
-
"grad_norm": 1.7401214838027954,
|
2787 |
-
"learning_rate": 2.925531914893617e-07,
|
2788 |
-
"loss": 2.173,
|
2789 |
-
"step": 396
|
2790 |
-
},
|
2791 |
-
{
|
2792 |
-
"epoch": 1.3327738145195132,
|
2793 |
-
"grad_norm": 0.5152906179428101,
|
2794 |
-
"learning_rate": 2.919621749408983e-07,
|
2795 |
-
"loss": 2.0993,
|
2796 |
-
"step": 397
|
2797 |
-
},
|
2798 |
-
{
|
2799 |
-
"epoch": 1.3361309274024338,
|
2800 |
-
"grad_norm": 0.5892696976661682,
|
2801 |
-
"learning_rate": 2.91371158392435e-07,
|
2802 |
-
"loss": 2.1427,
|
2803 |
-
"step": 398
|
2804 |
-
},
|
2805 |
-
{
|
2806 |
-
"epoch": 1.3394880402853546,
|
2807 |
-
"grad_norm": 0.5898533463478088,
|
2808 |
-
"learning_rate": 2.907801418439716e-07,
|
2809 |
-
"loss": 2.2104,
|
2810 |
-
"step": 399
|
2811 |
-
},
|
2812 |
-
{
|
2813 |
-
"epoch": 1.3428451531682752,
|
2814 |
-
"grad_norm": 0.5784264802932739,
|
2815 |
-
"learning_rate": 2.901891252955083e-07,
|
2816 |
-
"loss": 2.1015,
|
2817 |
-
"step": 400
|
2818 |
-
},
|
2819 |
-
{
|
2820 |
-
"epoch": 1.346202266051196,
|
2821 |
-
"grad_norm": 0.542095422744751,
|
2822 |
-
"learning_rate": 2.8959810874704486e-07,
|
2823 |
-
"loss": 2.1862,
|
2824 |
-
"step": 401
|
2825 |
-
},
|
2826 |
-
{
|
2827 |
-
"epoch": 1.3495593789341167,
|
2828 |
-
"grad_norm": 0.5719075202941895,
|
2829 |
-
"learning_rate": 2.8900709219858153e-07,
|
2830 |
-
"loss": 2.1248,
|
2831 |
-
"step": 402
|
2832 |
-
},
|
2833 |
-
{
|
2834 |
-
"epoch": 1.3529164918170373,
|
2835 |
-
"grad_norm": 0.5260282754898071,
|
2836 |
-
"learning_rate": 2.884160756501182e-07,
|
2837 |
-
"loss": 2.1521,
|
2838 |
-
"step": 403
|
2839 |
-
},
|
2840 |
-
{
|
2841 |
-
"epoch": 1.356273604699958,
|
2842 |
-
"grad_norm": 0.5233082175254822,
|
2843 |
-
"learning_rate": 2.8782505910165483e-07,
|
2844 |
-
"loss": 2.1747,
|
2845 |
-
"step": 404
|
2846 |
-
},
|
2847 |
-
{
|
2848 |
-
"epoch": 1.3596307175828788,
|
2849 |
-
"grad_norm": 0.6177014708518982,
|
2850 |
-
"learning_rate": 2.872340425531915e-07,
|
2851 |
-
"loss": 2.1494,
|
2852 |
-
"step": 405
|
2853 |
-
},
|
2854 |
-
{
|
2855 |
-
"epoch": 1.3629878304657994,
|
2856 |
-
"grad_norm": 0.5536350607872009,
|
2857 |
-
"learning_rate": 2.8664302600472813e-07,
|
2858 |
-
"loss": 2.1911,
|
2859 |
-
"step": 406
|
2860 |
-
},
|
2861 |
-
{
|
2862 |
-
"epoch": 1.36634494334872,
|
2863 |
-
"grad_norm": 0.550588071346283,
|
2864 |
-
"learning_rate": 2.860520094562648e-07,
|
2865 |
-
"loss": 2.1531,
|
2866 |
-
"step": 407
|
2867 |
-
},
|
2868 |
-
{
|
2869 |
-
"epoch": 1.3697020562316409,
|
2870 |
-
"grad_norm": 0.5359508395195007,
|
2871 |
-
"learning_rate": 2.854609929078014e-07,
|
2872 |
-
"loss": 2.1343,
|
2873 |
-
"step": 408
|
2874 |
-
},
|
2875 |
-
{
|
2876 |
-
"epoch": 1.3730591691145615,
|
2877 |
-
"grad_norm": 0.5293654799461365,
|
2878 |
-
"learning_rate": 2.8486997635933806e-07,
|
2879 |
-
"loss": 2.1612,
|
2880 |
-
"step": 409
|
2881 |
-
},
|
2882 |
-
{
|
2883 |
-
"epoch": 1.3764162819974821,
|
2884 |
-
"grad_norm": 0.5226705074310303,
|
2885 |
-
"learning_rate": 2.842789598108747e-07,
|
2886 |
-
"loss": 2.1633,
|
2887 |
-
"step": 410
|
2888 |
-
},
|
2889 |
-
{
|
2890 |
-
"epoch": 1.3797733948804027,
|
2891 |
-
"grad_norm": 0.5413585305213928,
|
2892 |
-
"learning_rate": 2.8368794326241136e-07,
|
2893 |
-
"loss": 2.1663,
|
2894 |
-
"step": 411
|
2895 |
-
},
|
2896 |
-
{
|
2897 |
-
"epoch": 1.3831305077633236,
|
2898 |
-
"grad_norm": 0.546417236328125,
|
2899 |
-
"learning_rate": 2.83096926713948e-07,
|
2900 |
-
"loss": 2.1514,
|
2901 |
-
"step": 412
|
2902 |
-
},
|
2903 |
-
{
|
2904 |
-
"epoch": 1.3864876206462442,
|
2905 |
-
"grad_norm": 0.5406803488731384,
|
2906 |
-
"learning_rate": 2.8250591016548466e-07,
|
2907 |
-
"loss": 2.209,
|
2908 |
-
"step": 413
|
2909 |
-
},
|
2910 |
-
{
|
2911 |
-
"epoch": 1.389844733529165,
|
2912 |
-
"grad_norm": 0.5303312540054321,
|
2913 |
-
"learning_rate": 2.8191489361702123e-07,
|
2914 |
-
"loss": 2.1523,
|
2915 |
-
"step": 414
|
2916 |
-
},
|
2917 |
-
{
|
2918 |
-
"epoch": 1.3932018464120857,
|
2919 |
-
"grad_norm": 0.5181916356086731,
|
2920 |
-
"learning_rate": 2.813238770685579e-07,
|
2921 |
-
"loss": 2.1744,
|
2922 |
-
"step": 415
|
2923 |
-
},
|
2924 |
-
{
|
2925 |
-
"epoch": 1.3965589592950063,
|
2926 |
-
"grad_norm": 0.5345380902290344,
|
2927 |
-
"learning_rate": 2.8073286052009453e-07,
|
2928 |
-
"loss": 2.1623,
|
2929 |
-
"step": 416
|
2930 |
-
},
|
2931 |
-
{
|
2932 |
-
"epoch": 1.399916072177927,
|
2933 |
-
"grad_norm": 0.5157522559165955,
|
2934 |
-
"learning_rate": 2.801418439716312e-07,
|
2935 |
-
"loss": 2.1714,
|
2936 |
-
"step": 417
|
2937 |
-
},
|
2938 |
-
{
|
2939 |
-
"epoch": 1.4032731850608477,
|
2940 |
-
"grad_norm": 0.5216273069381714,
|
2941 |
-
"learning_rate": 2.7955082742316783e-07,
|
2942 |
-
"loss": 2.1456,
|
2943 |
-
"step": 418
|
2944 |
-
},
|
2945 |
-
{
|
2946 |
-
"epoch": 1.4066302979437684,
|
2947 |
-
"grad_norm": 0.5333211421966553,
|
2948 |
-
"learning_rate": 2.789598108747045e-07,
|
2949 |
-
"loss": 2.1617,
|
2950 |
-
"step": 419
|
2951 |
-
},
|
2952 |
-
{
|
2953 |
-
"epoch": 1.409987410826689,
|
2954 |
-
"grad_norm": 0.5298429131507874,
|
2955 |
-
"learning_rate": 2.783687943262411e-07,
|
2956 |
-
"loss": 2.1636,
|
2957 |
-
"step": 420
|
2958 |
-
},
|
2959 |
-
{
|
2960 |
-
"epoch": 1.4133445237096098,
|
2961 |
-
"grad_norm": 0.5449317693710327,
|
2962 |
-
"learning_rate": 2.7777777777777776e-07,
|
2963 |
-
"loss": 2.1789,
|
2964 |
-
"step": 421
|
2965 |
-
},
|
2966 |
-
{
|
2967 |
-
"epoch": 1.4167016365925305,
|
2968 |
-
"grad_norm": 0.57036954164505,
|
2969 |
-
"learning_rate": 2.771867612293144e-07,
|
2970 |
-
"loss": 2.0884,
|
2971 |
-
"step": 422
|
2972 |
-
},
|
2973 |
-
{
|
2974 |
-
"epoch": 1.420058749475451,
|
2975 |
-
"grad_norm": 0.5396662950515747,
|
2976 |
-
"learning_rate": 2.7659574468085106e-07,
|
2977 |
-
"loss": 2.1662,
|
2978 |
-
"step": 423
|
2979 |
-
},
|
2980 |
-
{
|
2981 |
-
"epoch": 1.4234158623583717,
|
2982 |
-
"grad_norm": 0.5240216255187988,
|
2983 |
-
"learning_rate": 2.760047281323877e-07,
|
2984 |
-
"loss": 2.1325,
|
2985 |
-
"step": 424
|
2986 |
-
},
|
2987 |
-
{
|
2988 |
-
"epoch": 1.4267729752412925,
|
2989 |
-
"grad_norm": 0.5430148243904114,
|
2990 |
-
"learning_rate": 2.7541371158392436e-07,
|
2991 |
-
"loss": 2.1475,
|
2992 |
-
"step": 425
|
2993 |
-
},
|
2994 |
-
{
|
2995 |
-
"epoch": 1.4301300881242132,
|
2996 |
-
"grad_norm": 0.503663957118988,
|
2997 |
-
"learning_rate": 2.74822695035461e-07,
|
2998 |
-
"loss": 2.112,
|
2999 |
-
"step": 426
|
3000 |
-
},
|
3001 |
-
{
|
3002 |
-
"epoch": 1.4334872010071338,
|
3003 |
-
"grad_norm": 0.5414028167724609,
|
3004 |
-
"learning_rate": 2.742316784869976e-07,
|
3005 |
-
"loss": 2.1421,
|
3006 |
-
"step": 427
|
3007 |
-
},
|
3008 |
-
{
|
3009 |
-
"epoch": 1.4368443138900546,
|
3010 |
-
"grad_norm": 0.5510504841804504,
|
3011 |
-
"learning_rate": 2.7364066193853423e-07,
|
3012 |
-
"loss": 2.1442,
|
3013 |
-
"step": 428
|
3014 |
-
},
|
3015 |
-
{
|
3016 |
-
"epoch": 1.4402014267729752,
|
3017 |
-
"grad_norm": 0.5682281851768494,
|
3018 |
-
"learning_rate": 2.730496453900709e-07,
|
3019 |
-
"loss": 2.1771,
|
3020 |
-
"step": 429
|
3021 |
-
},
|
3022 |
-
{
|
3023 |
-
"epoch": 1.4435585396558959,
|
3024 |
-
"grad_norm": 0.5418784618377686,
|
3025 |
-
"learning_rate": 2.7245862884160753e-07,
|
3026 |
-
"loss": 2.1669,
|
3027 |
-
"step": 430
|
3028 |
-
},
|
3029 |
-
{
|
3030 |
-
"epoch": 1.4469156525388165,
|
3031 |
-
"grad_norm": 0.5360921025276184,
|
3032 |
-
"learning_rate": 2.718676122931442e-07,
|
3033 |
-
"loss": 2.1464,
|
3034 |
-
"step": 431
|
3035 |
-
},
|
3036 |
-
{
|
3037 |
-
"epoch": 1.4502727654217373,
|
3038 |
-
"grad_norm": 0.5701574683189392,
|
3039 |
-
"learning_rate": 2.7127659574468084e-07,
|
3040 |
-
"loss": 2.1822,
|
3041 |
-
"step": 432
|
3042 |
-
},
|
3043 |
-
{
|
3044 |
-
"epoch": 1.453629878304658,
|
3045 |
-
"grad_norm": 0.5453020930290222,
|
3046 |
-
"learning_rate": 2.7068557919621746e-07,
|
3047 |
-
"loss": 2.1729,
|
3048 |
-
"step": 433
|
3049 |
-
},
|
3050 |
-
{
|
3051 |
-
"epoch": 1.4569869911875788,
|
3052 |
-
"grad_norm": 0.5435007810592651,
|
3053 |
-
"learning_rate": 2.700945626477541e-07,
|
3054 |
-
"loss": 2.1511,
|
3055 |
-
"step": 434
|
3056 |
-
},
|
3057 |
-
{
|
3058 |
-
"epoch": 1.4603441040704994,
|
3059 |
-
"grad_norm": 0.5011966228485107,
|
3060 |
-
"learning_rate": 2.6950354609929076e-07,
|
3061 |
-
"loss": 2.108,
|
3062 |
-
"step": 435
|
3063 |
-
},
|
3064 |
-
{
|
3065 |
-
"epoch": 1.46370121695342,
|
3066 |
-
"grad_norm": 0.49755868315696716,
|
3067 |
-
"learning_rate": 2.6891252955082744e-07,
|
3068 |
-
"loss": 2.1776,
|
3069 |
-
"step": 436
|
3070 |
-
},
|
3071 |
-
{
|
3072 |
-
"epoch": 1.4670583298363407,
|
3073 |
-
"grad_norm": 0.5629771947860718,
|
3074 |
-
"learning_rate": 2.6832151300236406e-07,
|
3075 |
-
"loss": 2.1493,
|
3076 |
-
"step": 437
|
3077 |
-
},
|
3078 |
-
{
|
3079 |
-
"epoch": 1.4704154427192615,
|
3080 |
-
"grad_norm": 0.5360575914382935,
|
3081 |
-
"learning_rate": 2.6773049645390074e-07,
|
3082 |
-
"loss": 2.1732,
|
3083 |
-
"step": 438
|
3084 |
-
},
|
3085 |
-
{
|
3086 |
-
"epoch": 1.4737725556021821,
|
3087 |
-
"grad_norm": 0.5516806840896606,
|
3088 |
-
"learning_rate": 2.6713947990543736e-07,
|
3089 |
-
"loss": 2.148,
|
3090 |
-
"step": 439
|
3091 |
-
},
|
3092 |
-
{
|
3093 |
-
"epoch": 1.4771296684851027,
|
3094 |
-
"grad_norm": 0.5595859289169312,
|
3095 |
-
"learning_rate": 2.66548463356974e-07,
|
3096 |
-
"loss": 2.1671,
|
3097 |
-
"step": 440
|
3098 |
-
},
|
3099 |
-
{
|
3100 |
-
"epoch": 1.4804867813680236,
|
3101 |
-
"grad_norm": 0.5414209365844727,
|
3102 |
-
"learning_rate": 2.659574468085106e-07,
|
3103 |
-
"loss": 2.1233,
|
3104 |
-
"step": 441
|
3105 |
-
},
|
3106 |
-
{
|
3107 |
-
"epoch": 1.4838438942509442,
|
3108 |
-
"grad_norm": 0.5354856252670288,
|
3109 |
-
"learning_rate": 2.653664302600473e-07,
|
3110 |
-
"loss": 2.1921,
|
3111 |
-
"step": 442
|
3112 |
-
},
|
3113 |
-
{
|
3114 |
-
"epoch": 1.4872010071338648,
|
3115 |
-
"grad_norm": 0.565666139125824,
|
3116 |
-
"learning_rate": 2.647754137115839e-07,
|
3117 |
-
"loss": 2.198,
|
3118 |
-
"step": 443
|
3119 |
-
},
|
3120 |
-
{
|
3121 |
-
"epoch": 1.4905581200167854,
|
3122 |
-
"grad_norm": 0.5538352727890015,
|
3123 |
-
"learning_rate": 2.641843971631206e-07,
|
3124 |
-
"loss": 2.1396,
|
3125 |
-
"step": 444
|
3126 |
-
},
|
3127 |
-
{
|
3128 |
-
"epoch": 1.4939152328997063,
|
3129 |
-
"grad_norm": 0.536029577255249,
|
3130 |
-
"learning_rate": 2.635933806146572e-07,
|
3131 |
-
"loss": 2.1811,
|
3132 |
-
"step": 445
|
3133 |
-
},
|
3134 |
-
{
|
3135 |
-
"epoch": 1.497272345782627,
|
3136 |
-
"grad_norm": 0.5602884888648987,
|
3137 |
-
"learning_rate": 2.6300236406619384e-07,
|
3138 |
-
"loss": 2.1429,
|
3139 |
-
"step": 446
|
3140 |
-
},
|
3141 |
-
{
|
3142 |
-
"epoch": 1.5006294586655478,
|
3143 |
-
"grad_norm": 0.5486292243003845,
|
3144 |
-
"learning_rate": 2.6241134751773046e-07,
|
3145 |
-
"loss": 2.1153,
|
3146 |
-
"step": 447
|
3147 |
-
},
|
3148 |
-
{
|
3149 |
-
"epoch": 1.5039865715484684,
|
3150 |
-
"grad_norm": 0.5302276611328125,
|
3151 |
-
"learning_rate": 2.6182033096926714e-07,
|
3152 |
-
"loss": 2.1709,
|
3153 |
-
"step": 448
|
3154 |
-
},
|
3155 |
-
{
|
3156 |
-
"epoch": 1.507343684431389,
|
3157 |
-
"grad_norm": 0.5530226230621338,
|
3158 |
-
"learning_rate": 2.6122931442080376e-07,
|
3159 |
-
"loss": 2.1891,
|
3160 |
-
"step": 449
|
3161 |
-
},
|
3162 |
-
{
|
3163 |
-
"epoch": 1.5107007973143096,
|
3164 |
-
"grad_norm": 0.5639198422431946,
|
3165 |
-
"learning_rate": 2.6063829787234044e-07,
|
3166 |
-
"loss": 2.1415,
|
3167 |
-
"step": 450
|
3168 |
-
},
|
3169 |
-
{
|
3170 |
-
"epoch": 1.5140579101972302,
|
3171 |
-
"grad_norm": 0.5313774943351746,
|
3172 |
-
"learning_rate": 2.6004728132387706e-07,
|
3173 |
-
"loss": 2.117,
|
3174 |
-
"step": 451
|
3175 |
-
},
|
3176 |
-
{
|
3177 |
-
"epoch": 1.517415023080151,
|
3178 |
-
"grad_norm": 0.5192495584487915,
|
3179 |
-
"learning_rate": 2.5945626477541374e-07,
|
3180 |
-
"loss": 2.1827,
|
3181 |
-
"step": 452
|
3182 |
-
},
|
3183 |
-
{
|
3184 |
-
"epoch": 1.5207721359630717,
|
3185 |
-
"grad_norm": 0.5575863718986511,
|
3186 |
-
"learning_rate": 2.588652482269503e-07,
|
3187 |
-
"loss": 2.1635,
|
3188 |
-
"step": 453
|
3189 |
-
},
|
3190 |
-
{
|
3191 |
-
"epoch": 1.5241292488459925,
|
3192 |
-
"grad_norm": 0.569666862487793,
|
3193 |
-
"learning_rate": 2.58274231678487e-07,
|
3194 |
-
"loss": 2.1455,
|
3195 |
-
"step": 454
|
3196 |
-
},
|
3197 |
-
{
|
3198 |
-
"epoch": 1.5274863617289132,
|
3199 |
-
"grad_norm": 0.5995537042617798,
|
3200 |
-
"learning_rate": 2.576832151300236e-07,
|
3201 |
-
"loss": 2.1799,
|
3202 |
-
"step": 455
|
3203 |
-
},
|
3204 |
-
{
|
3205 |
-
"epoch": 1.5308434746118338,
|
3206 |
-
"grad_norm": 0.5367946624755859,
|
3207 |
-
"learning_rate": 2.570921985815603e-07,
|
3208 |
-
"loss": 2.1031,
|
3209 |
-
"step": 456
|
3210 |
-
},
|
3211 |
-
{
|
3212 |
-
"epoch": 1.5342005874947544,
|
3213 |
-
"grad_norm": 0.5633010864257812,
|
3214 |
-
"learning_rate": 2.565011820330969e-07,
|
3215 |
-
"loss": 2.1116,
|
3216 |
-
"step": 457
|
3217 |
-
},
|
3218 |
-
{
|
3219 |
-
"epoch": 1.5375577003776753,
|
3220 |
-
"grad_norm": 0.5201706886291504,
|
3221 |
-
"learning_rate": 2.559101654846336e-07,
|
3222 |
-
"loss": 2.1741,
|
3223 |
-
"step": 458
|
3224 |
-
},
|
3225 |
-
{
|
3226 |
-
"epoch": 1.5409148132605959,
|
3227 |
-
"grad_norm": 0.5941653251647949,
|
3228 |
-
"learning_rate": 2.5531914893617016e-07,
|
3229 |
-
"loss": 2.2537,
|
3230 |
-
"step": 459
|
3231 |
-
},
|
3232 |
-
{
|
3233 |
-
"epoch": 1.5442719261435167,
|
3234 |
-
"grad_norm": 0.5953680872917175,
|
3235 |
-
"learning_rate": 2.5472813238770684e-07,
|
3236 |
-
"loss": 2.1095,
|
3237 |
-
"step": 460
|
3238 |
-
},
|
3239 |
-
{
|
3240 |
-
"epoch": 1.5476290390264373,
|
3241 |
-
"grad_norm": 0.5477280020713806,
|
3242 |
-
"learning_rate": 2.5413711583924346e-07,
|
3243 |
-
"loss": 2.1789,
|
3244 |
-
"step": 461
|
3245 |
-
},
|
3246 |
-
{
|
3247 |
-
"epoch": 1.550986151909358,
|
3248 |
-
"grad_norm": 0.5424633622169495,
|
3249 |
-
"learning_rate": 2.5354609929078014e-07,
|
3250 |
-
"loss": 2.1394,
|
3251 |
-
"step": 462
|
3252 |
-
},
|
3253 |
-
{
|
3254 |
-
"epoch": 1.5543432647922786,
|
3255 |
-
"grad_norm": 0.5671295523643494,
|
3256 |
-
"learning_rate": 2.5295508274231676e-07,
|
3257 |
-
"loss": 2.159,
|
3258 |
-
"step": 463
|
3259 |
-
},
|
3260 |
-
{
|
3261 |
-
"epoch": 1.5577003776751992,
|
3262 |
-
"grad_norm": 0.5310060977935791,
|
3263 |
-
"learning_rate": 2.5236406619385344e-07,
|
3264 |
-
"loss": 2.1596,
|
3265 |
-
"step": 464
|
3266 |
-
},
|
3267 |
-
{
|
3268 |
-
"epoch": 1.56105749055812,
|
3269 |
-
"grad_norm": 0.5472586154937744,
|
3270 |
-
"learning_rate": 2.5177304964539e-07,
|
3271 |
-
"loss": 2.1166,
|
3272 |
-
"step": 465
|
3273 |
-
},
|
3274 |
-
{
|
3275 |
-
"epoch": 1.5644146034410407,
|
3276 |
-
"grad_norm": 0.5335962176322937,
|
3277 |
-
"learning_rate": 2.511820330969267e-07,
|
3278 |
-
"loss": 2.1427,
|
3279 |
-
"step": 466
|
3280 |
-
},
|
3281 |
-
{
|
3282 |
-
"epoch": 1.5677717163239615,
|
3283 |
-
"grad_norm": 0.5123631954193115,
|
3284 |
-
"learning_rate": 2.505910165484633e-07,
|
3285 |
-
"loss": 2.0675,
|
3286 |
-
"step": 467
|
3287 |
-
},
|
3288 |
-
{
|
3289 |
-
"epoch": 1.5711288292068821,
|
3290 |
-
"grad_norm": 0.5876572728157043,
|
3291 |
-
"learning_rate": 2.5e-07,
|
3292 |
-
"loss": 2.1258,
|
3293 |
-
"step": 468
|
3294 |
-
},
|
3295 |
-
{
|
3296 |
-
"epoch": 1.5744859420898027,
|
3297 |
-
"grad_norm": 0.5395733118057251,
|
3298 |
-
"learning_rate": 2.494089834515366e-07,
|
3299 |
-
"loss": 2.1389,
|
3300 |
-
"step": 469
|
3301 |
-
},
|
3302 |
-
{
|
3303 |
-
"epoch": 1.5778430549727234,
|
3304 |
-
"grad_norm": 0.5670069456100464,
|
3305 |
-
"learning_rate": 2.488179669030733e-07,
|
3306 |
-
"loss": 2.2214,
|
3307 |
-
"step": 470
|
3308 |
-
},
|
3309 |
-
{
|
3310 |
-
"epoch": 1.581200167855644,
|
3311 |
-
"grad_norm": 0.5086778998374939,
|
3312 |
-
"learning_rate": 2.482269503546099e-07,
|
3313 |
-
"loss": 2.1135,
|
3314 |
-
"step": 471
|
3315 |
-
},
|
3316 |
-
{
|
3317 |
-
"epoch": 1.5845572807385648,
|
3318 |
-
"grad_norm": 0.5723012089729309,
|
3319 |
-
"learning_rate": 2.4763593380614654e-07,
|
3320 |
-
"loss": 2.2149,
|
3321 |
-
"step": 472
|
3322 |
-
},
|
3323 |
-
{
|
3324 |
-
"epoch": 1.5879143936214857,
|
3325 |
-
"grad_norm": 0.5220308303833008,
|
3326 |
-
"learning_rate": 2.470449172576832e-07,
|
3327 |
-
"loss": 2.1794,
|
3328 |
-
"step": 473
|
3329 |
-
},
|
3330 |
-
{
|
3331 |
-
"epoch": 1.5912715065044063,
|
3332 |
-
"grad_norm": 0.5964010953903198,
|
3333 |
-
"learning_rate": 2.4645390070921984e-07,
|
3334 |
-
"loss": 2.2251,
|
3335 |
-
"step": 474
|
3336 |
-
},
|
3337 |
-
{
|
3338 |
-
"epoch": 1.594628619387327,
|
3339 |
-
"grad_norm": 0.5133795738220215,
|
3340 |
-
"learning_rate": 2.4586288416075646e-07,
|
3341 |
-
"loss": 2.1747,
|
3342 |
-
"step": 475
|
3343 |
-
},
|
3344 |
-
{
|
3345 |
-
"epoch": 1.5979857322702475,
|
3346 |
-
"grad_norm": 0.5728108286857605,
|
3347 |
-
"learning_rate": 2.4527186761229314e-07,
|
3348 |
-
"loss": 2.2154,
|
3349 |
-
"step": 476
|
3350 |
-
},
|
3351 |
-
{
|
3352 |
-
"epoch": 1.6013428451531682,
|
3353 |
-
"grad_norm": 0.5136426687240601,
|
3354 |
-
"learning_rate": 2.4468085106382976e-07,
|
3355 |
-
"loss": 2.146,
|
3356 |
-
"step": 477
|
3357 |
-
},
|
3358 |
-
{
|
3359 |
-
"epoch": 1.604699958036089,
|
3360 |
-
"grad_norm": 0.5617910027503967,
|
3361 |
-
"learning_rate": 2.440898345153664e-07,
|
3362 |
-
"loss": 2.1504,
|
3363 |
-
"step": 478
|
3364 |
-
},
|
3365 |
-
{
|
3366 |
-
"epoch": 1.6080570709190096,
|
3367 |
-
"grad_norm": 0.5407204031944275,
|
3368 |
-
"learning_rate": 2.4349881796690306e-07,
|
3369 |
-
"loss": 2.1544,
|
3370 |
-
"step": 479
|
3371 |
-
},
|
3372 |
-
{
|
3373 |
-
"epoch": 1.6114141838019305,
|
3374 |
-
"grad_norm": 0.5309438109397888,
|
3375 |
-
"learning_rate": 2.429078014184397e-07,
|
3376 |
-
"loss": 2.128,
|
3377 |
-
"step": 480
|
3378 |
-
},
|
3379 |
-
{
|
3380 |
-
"epoch": 1.614771296684851,
|
3381 |
-
"grad_norm": 0.5255685448646545,
|
3382 |
-
"learning_rate": 2.423167848699763e-07,
|
3383 |
-
"loss": 2.0676,
|
3384 |
-
"step": 481
|
3385 |
-
},
|
3386 |
-
{
|
3387 |
-
"epoch": 1.6181284095677717,
|
3388 |
-
"grad_norm": 0.5453495383262634,
|
3389 |
-
"learning_rate": 2.41725768321513e-07,
|
3390 |
-
"loss": 2.1854,
|
3391 |
-
"step": 482
|
3392 |
-
},
|
3393 |
-
{
|
3394 |
-
"epoch": 1.6214855224506923,
|
3395 |
-
"grad_norm": 0.5266616940498352,
|
3396 |
-
"learning_rate": 2.411347517730496e-07,
|
3397 |
-
"loss": 2.145,
|
3398 |
-
"step": 483
|
3399 |
-
},
|
3400 |
-
{
|
3401 |
-
"epoch": 1.624842635333613,
|
3402 |
-
"grad_norm": 0.5247876644134521,
|
3403 |
-
"learning_rate": 2.4054373522458624e-07,
|
3404 |
-
"loss": 2.1812,
|
3405 |
-
"step": 484
|
3406 |
-
},
|
3407 |
-
{
|
3408 |
-
"epoch": 1.6281997482165338,
|
3409 |
-
"grad_norm": 0.5246554613113403,
|
3410 |
-
"learning_rate": 2.399527186761229e-07,
|
3411 |
-
"loss": 2.1136,
|
3412 |
-
"step": 485
|
3413 |
-
},
|
3414 |
-
{
|
3415 |
-
"epoch": 1.6315568610994544,
|
3416 |
-
"grad_norm": 0.5595040917396545,
|
3417 |
-
"learning_rate": 2.393617021276596e-07,
|
3418 |
-
"loss": 2.1728,
|
3419 |
-
"step": 486
|
3420 |
-
},
|
3421 |
-
{
|
3422 |
-
"epoch": 1.6349139739823753,
|
3423 |
-
"grad_norm": 0.533994197845459,
|
3424 |
-
"learning_rate": 2.387706855791962e-07,
|
3425 |
-
"loss": 2.0939,
|
3426 |
-
"step": 487
|
3427 |
-
},
|
3428 |
-
{
|
3429 |
-
"epoch": 1.6382710868652959,
|
3430 |
-
"grad_norm": 0.5742591023445129,
|
3431 |
-
"learning_rate": 2.3817966903073287e-07,
|
3432 |
-
"loss": 2.148,
|
3433 |
-
"step": 488
|
3434 |
-
},
|
3435 |
-
{
|
3436 |
-
"epoch": 1.6416281997482165,
|
3437 |
-
"grad_norm": 0.5369508266448975,
|
3438 |
-
"learning_rate": 2.375886524822695e-07,
|
3439 |
-
"loss": 2.1524,
|
3440 |
-
"step": 489
|
3441 |
-
},
|
3442 |
-
{
|
3443 |
-
"epoch": 1.6449853126311371,
|
3444 |
-
"grad_norm": 0.5108909010887146,
|
3445 |
-
"learning_rate": 2.3699763593380614e-07,
|
3446 |
-
"loss": 2.1586,
|
3447 |
-
"step": 490
|
3448 |
-
},
|
3449 |
-
{
|
3450 |
-
"epoch": 1.648342425514058,
|
3451 |
-
"grad_norm": 0.5841718912124634,
|
3452 |
-
"learning_rate": 2.364066193853428e-07,
|
3453 |
-
"loss": 2.1854,
|
3454 |
-
"step": 491
|
3455 |
-
},
|
3456 |
-
{
|
3457 |
-
"epoch": 1.6516995383969786,
|
3458 |
-
"grad_norm": 0.6200852394104004,
|
3459 |
-
"learning_rate": 2.3581560283687944e-07,
|
3460 |
-
"loss": 2.1811,
|
3461 |
-
"step": 492
|
3462 |
-
},
|
3463 |
-
{
|
3464 |
-
"epoch": 1.6550566512798994,
|
3465 |
-
"grad_norm": 0.5158974528312683,
|
3466 |
-
"learning_rate": 2.3522458628841607e-07,
|
3467 |
-
"loss": 2.1568,
|
3468 |
-
"step": 493
|
3469 |
-
},
|
3470 |
-
{
|
3471 |
-
"epoch": 1.65841376416282,
|
3472 |
-
"grad_norm": 0.5737884044647217,
|
3473 |
-
"learning_rate": 2.3463356973995272e-07,
|
3474 |
-
"loss": 2.1639,
|
3475 |
-
"step": 494
|
3476 |
-
},
|
3477 |
-
{
|
3478 |
-
"epoch": 1.6617708770457407,
|
3479 |
-
"grad_norm": 0.5661967396736145,
|
3480 |
-
"learning_rate": 2.3404255319148937e-07,
|
3481 |
-
"loss": 2.1164,
|
3482 |
-
"step": 495
|
3483 |
-
},
|
3484 |
-
{
|
3485 |
-
"epoch": 1.6651279899286613,
|
3486 |
-
"grad_norm": 0.5981073379516602,
|
3487 |
-
"learning_rate": 2.33451536643026e-07,
|
3488 |
-
"loss": 2.1875,
|
3489 |
-
"step": 496
|
3490 |
-
},
|
3491 |
-
{
|
3492 |
-
"epoch": 1.668485102811582,
|
3493 |
-
"grad_norm": 0.5757937431335449,
|
3494 |
-
"learning_rate": 2.3286052009456264e-07,
|
3495 |
-
"loss": 2.1315,
|
3496 |
-
"step": 497
|
3497 |
-
},
|
3498 |
-
{
|
3499 |
-
"epoch": 1.6718422156945028,
|
3500 |
-
"grad_norm": 0.5519205927848816,
|
3501 |
-
"learning_rate": 2.322695035460993e-07,
|
3502 |
-
"loss": 2.1126,
|
3503 |
-
"step": 498
|
3504 |
-
},
|
3505 |
-
{
|
3506 |
-
"epoch": 1.6751993285774234,
|
3507 |
-
"grad_norm": 0.5893654823303223,
|
3508 |
-
"learning_rate": 2.3167848699763592e-07,
|
3509 |
-
"loss": 2.1427,
|
3510 |
-
"step": 499
|
3511 |
-
},
|
3512 |
-
{
|
3513 |
-
"epoch": 1.6785564414603442,
|
3514 |
-
"grad_norm": 0.5640158653259277,
|
3515 |
-
"learning_rate": 2.3108747044917257e-07,
|
3516 |
-
"loss": 2.1763,
|
3517 |
-
"step": 500
|
3518 |
-
},
|
3519 |
-
{
|
3520 |
-
"epoch": 1.6819135543432648,
|
3521 |
-
"grad_norm": 0.5426249504089355,
|
3522 |
-
"learning_rate": 2.3049645390070922e-07,
|
3523 |
-
"loss": 2.0914,
|
3524 |
-
"step": 501
|
3525 |
-
},
|
3526 |
-
{
|
3527 |
-
"epoch": 1.6852706672261855,
|
3528 |
-
"grad_norm": 0.5572645664215088,
|
3529 |
-
"learning_rate": 2.2990543735224584e-07,
|
3530 |
-
"loss": 2.1482,
|
3531 |
-
"step": 502
|
3532 |
-
},
|
3533 |
-
{
|
3534 |
-
"epoch": 1.688627780109106,
|
3535 |
-
"grad_norm": 0.5625807642936707,
|
3536 |
-
"learning_rate": 2.293144208037825e-07,
|
3537 |
-
"loss": 2.1804,
|
3538 |
-
"step": 503
|
3539 |
-
},
|
3540 |
-
{
|
3541 |
-
"epoch": 1.6919848929920267,
|
3542 |
-
"grad_norm": 0.56282639503479,
|
3543 |
-
"learning_rate": 2.2872340425531914e-07,
|
3544 |
-
"loss": 2.1613,
|
3545 |
-
"step": 504
|
3546 |
-
},
|
3547 |
-
{
|
3548 |
-
"epoch": 1.6953420058749475,
|
3549 |
-
"grad_norm": 0.5359474420547485,
|
3550 |
-
"learning_rate": 2.2813238770685577e-07,
|
3551 |
-
"loss": 2.1019,
|
3552 |
-
"step": 505
|
3553 |
-
},
|
3554 |
-
{
|
3555 |
-
"epoch": 1.6986991187578684,
|
3556 |
-
"grad_norm": 0.5989973545074463,
|
3557 |
-
"learning_rate": 2.2754137115839242e-07,
|
3558 |
-
"loss": 2.2191,
|
3559 |
-
"step": 506
|
3560 |
-
},
|
3561 |
-
{
|
3562 |
-
"epoch": 1.702056231640789,
|
3563 |
-
"grad_norm": 0.5682387351989746,
|
3564 |
-
"learning_rate": 2.2695035460992907e-07,
|
3565 |
-
"loss": 2.1351,
|
3566 |
-
"step": 507
|
3567 |
-
},
|
3568 |
-
{
|
3569 |
-
"epoch": 1.7054133445237096,
|
3570 |
-
"grad_norm": 0.5352203845977783,
|
3571 |
-
"learning_rate": 2.2635933806146572e-07,
|
3572 |
-
"loss": 2.1098,
|
3573 |
-
"step": 508
|
3574 |
-
},
|
3575 |
-
{
|
3576 |
-
"epoch": 1.7087704574066302,
|
3577 |
-
"grad_norm": 0.5701361894607544,
|
3578 |
-
"learning_rate": 2.2576832151300234e-07,
|
3579 |
-
"loss": 2.163,
|
3580 |
-
"step": 509
|
3581 |
-
},
|
3582 |
-
{
|
3583 |
-
"epoch": 1.7121275702895509,
|
3584 |
-
"grad_norm": 0.5832070112228394,
|
3585 |
-
"learning_rate": 2.25177304964539e-07,
|
3586 |
-
"loss": 2.1448,
|
3587 |
-
"step": 510
|
3588 |
-
},
|
3589 |
-
{
|
3590 |
-
"epoch": 1.7154846831724717,
|
3591 |
-
"grad_norm": 0.5940104722976685,
|
3592 |
-
"learning_rate": 2.2458628841607564e-07,
|
3593 |
-
"loss": 2.1963,
|
3594 |
-
"step": 511
|
3595 |
-
},
|
3596 |
-
{
|
3597 |
-
"epoch": 1.7188417960553923,
|
3598 |
-
"grad_norm": 0.5954880714416504,
|
3599 |
-
"learning_rate": 2.2399527186761227e-07,
|
3600 |
-
"loss": 2.1871,
|
3601 |
-
"step": 512
|
3602 |
-
},
|
3603 |
-
{
|
3604 |
-
"epoch": 1.7221989089383132,
|
3605 |
-
"grad_norm": 0.5460233688354492,
|
3606 |
-
"learning_rate": 2.2340425531914892e-07,
|
3607 |
-
"loss": 2.1694,
|
3608 |
-
"step": 513
|
3609 |
-
},
|
3610 |
-
{
|
3611 |
-
"epoch": 1.7255560218212338,
|
3612 |
-
"grad_norm": 0.5475773215293884,
|
3613 |
-
"learning_rate": 2.2281323877068557e-07,
|
3614 |
-
"loss": 2.1602,
|
3615 |
-
"step": 514
|
3616 |
-
},
|
3617 |
-
{
|
3618 |
-
"epoch": 1.7289131347041544,
|
3619 |
-
"grad_norm": 0.5809094905853271,
|
3620 |
-
"learning_rate": 2.222222222222222e-07,
|
3621 |
-
"loss": 2.1808,
|
3622 |
-
"step": 515
|
3623 |
-
},
|
3624 |
-
{
|
3625 |
-
"epoch": 1.732270247587075,
|
3626 |
-
"grad_norm": 0.5999802947044373,
|
3627 |
-
"learning_rate": 2.2163120567375884e-07,
|
3628 |
-
"loss": 2.1899,
|
3629 |
-
"step": 516
|
3630 |
-
},
|
3631 |
-
{
|
3632 |
-
"epoch": 1.7356273604699957,
|
3633 |
-
"grad_norm": 0.5715625882148743,
|
3634 |
-
"learning_rate": 2.2104018912529552e-07,
|
3635 |
-
"loss": 2.0947,
|
3636 |
-
"step": 517
|
3637 |
-
},
|
3638 |
-
{
|
3639 |
-
"epoch": 1.7389844733529165,
|
3640 |
-
"grad_norm": 0.5752885341644287,
|
3641 |
-
"learning_rate": 2.2044917257683214e-07,
|
3642 |
-
"loss": 2.1653,
|
3643 |
-
"step": 518
|
3644 |
-
},
|
3645 |
-
{
|
3646 |
-
"epoch": 1.7423415862358371,
|
3647 |
-
"grad_norm": 0.5773462057113647,
|
3648 |
-
"learning_rate": 2.198581560283688e-07,
|
3649 |
-
"loss": 2.0973,
|
3650 |
-
"step": 519
|
3651 |
-
},
|
3652 |
-
{
|
3653 |
-
"epoch": 1.745698699118758,
|
3654 |
-
"grad_norm": 0.528805136680603,
|
3655 |
-
"learning_rate": 2.1926713947990544e-07,
|
3656 |
-
"loss": 2.1449,
|
3657 |
-
"step": 520
|
3658 |
-
},
|
3659 |
-
{
|
3660 |
-
"epoch": 1.7490558120016786,
|
3661 |
-
"grad_norm": 0.5383672714233398,
|
3662 |
-
"learning_rate": 2.1867612293144207e-07,
|
3663 |
-
"loss": 2.153,
|
3664 |
-
"step": 521
|
3665 |
-
},
|
3666 |
-
{
|
3667 |
-
"epoch": 1.7524129248845992,
|
3668 |
-
"grad_norm": 0.534181535243988,
|
3669 |
-
"learning_rate": 2.1808510638297872e-07,
|
3670 |
-
"loss": 2.1449,
|
3671 |
-
"step": 522
|
3672 |
-
},
|
3673 |
-
{
|
3674 |
-
"epoch": 1.7557700377675198,
|
3675 |
-
"grad_norm": 0.5447178483009338,
|
3676 |
-
"learning_rate": 2.1749408983451537e-07,
|
3677 |
-
"loss": 2.1375,
|
3678 |
-
"step": 523
|
3679 |
-
},
|
3680 |
-
{
|
3681 |
-
"epoch": 1.7591271506504407,
|
3682 |
-
"grad_norm": 0.5227280855178833,
|
3683 |
-
"learning_rate": 2.1690307328605202e-07,
|
3684 |
-
"loss": 2.1,
|
3685 |
-
"step": 524
|
3686 |
-
},
|
3687 |
-
{
|
3688 |
-
"epoch": 1.7624842635333613,
|
3689 |
-
"grad_norm": 0.579774796962738,
|
3690 |
-
"learning_rate": 2.1631205673758864e-07,
|
3691 |
-
"loss": 2.1619,
|
3692 |
-
"step": 525
|
3693 |
-
},
|
3694 |
-
{
|
3695 |
-
"epoch": 1.7658413764162821,
|
3696 |
-
"grad_norm": 0.542693018913269,
|
3697 |
-
"learning_rate": 2.157210401891253e-07,
|
3698 |
-
"loss": 2.1568,
|
3699 |
-
"step": 526
|
3700 |
-
},
|
3701 |
-
{
|
3702 |
-
"epoch": 1.7691984892992028,
|
3703 |
-
"grad_norm": 0.5392715334892273,
|
3704 |
-
"learning_rate": 2.1513002364066194e-07,
|
3705 |
-
"loss": 2.0959,
|
3706 |
-
"step": 527
|
3707 |
-
},
|
3708 |
-
{
|
3709 |
-
"epoch": 1.7725556021821234,
|
3710 |
-
"grad_norm": 0.6038206219673157,
|
3711 |
-
"learning_rate": 2.1453900709219857e-07,
|
3712 |
-
"loss": 2.133,
|
3713 |
-
"step": 528
|
3714 |
-
},
|
3715 |
-
{
|
3716 |
-
"epoch": 1.775912715065044,
|
3717 |
-
"grad_norm": 0.5537463426589966,
|
3718 |
-
"learning_rate": 2.1394799054373522e-07,
|
3719 |
-
"loss": 2.0817,
|
3720 |
-
"step": 529
|
3721 |
-
},
|
3722 |
-
{
|
3723 |
-
"epoch": 1.7792698279479646,
|
3724 |
-
"grad_norm": 0.5747633576393127,
|
3725 |
-
"learning_rate": 2.1335697399527187e-07,
|
3726 |
-
"loss": 2.1342,
|
3727 |
-
"step": 530
|
3728 |
-
},
|
3729 |
-
{
|
3730 |
-
"epoch": 1.7826269408308855,
|
3731 |
-
"grad_norm": 0.5527968406677246,
|
3732 |
-
"learning_rate": 2.127659574468085e-07,
|
3733 |
-
"loss": 2.184,
|
3734 |
-
"step": 531
|
3735 |
-
},
|
3736 |
-
{
|
3737 |
-
"epoch": 1.785984053713806,
|
3738 |
-
"grad_norm": 0.5554253458976746,
|
3739 |
-
"learning_rate": 2.1217494089834514e-07,
|
3740 |
-
"loss": 2.1976,
|
3741 |
-
"step": 532
|
3742 |
-
},
|
3743 |
-
{
|
3744 |
-
"epoch": 1.789341166596727,
|
3745 |
-
"grad_norm": 0.5290229916572571,
|
3746 |
-
"learning_rate": 2.115839243498818e-07,
|
3747 |
-
"loss": 2.102,
|
3748 |
-
"step": 533
|
3749 |
-
},
|
3750 |
-
{
|
3751 |
-
"epoch": 1.7926982794796475,
|
3752 |
-
"grad_norm": 0.5803096890449524,
|
3753 |
-
"learning_rate": 2.1099290780141842e-07,
|
3754 |
-
"loss": 2.0815,
|
3755 |
-
"step": 534
|
3756 |
-
},
|
3757 |
-
{
|
3758 |
-
"epoch": 1.7960553923625682,
|
3759 |
-
"grad_norm": 0.5680854916572571,
|
3760 |
-
"learning_rate": 2.1040189125295507e-07,
|
3761 |
-
"loss": 2.1338,
|
3762 |
-
"step": 535
|
3763 |
-
},
|
3764 |
-
{
|
3765 |
-
"epoch": 1.7994125052454888,
|
3766 |
-
"grad_norm": 0.5866023302078247,
|
3767 |
-
"learning_rate": 2.0981087470449172e-07,
|
3768 |
-
"loss": 2.2191,
|
3769 |
-
"step": 536
|
3770 |
-
},
|
3771 |
-
{
|
3772 |
-
"epoch": 1.8027696181284094,
|
3773 |
-
"grad_norm": 0.5355091691017151,
|
3774 |
-
"learning_rate": 2.0921985815602834e-07,
|
3775 |
-
"loss": 2.1312,
|
3776 |
-
"step": 537
|
3777 |
-
},
|
3778 |
-
{
|
3779 |
-
"epoch": 1.8061267310113303,
|
3780 |
-
"grad_norm": 0.5850256681442261,
|
3781 |
-
"learning_rate": 2.08628841607565e-07,
|
3782 |
-
"loss": 2.1827,
|
3783 |
-
"step": 538
|
3784 |
-
},
|
3785 |
-
{
|
3786 |
-
"epoch": 1.809483843894251,
|
3787 |
-
"grad_norm": 0.55758136510849,
|
3788 |
-
"learning_rate": 2.0803782505910164e-07,
|
3789 |
-
"loss": 2.1193,
|
3790 |
-
"step": 539
|
3791 |
-
},
|
3792 |
-
{
|
3793 |
-
"epoch": 1.8128409567771717,
|
3794 |
-
"grad_norm": 0.5220862627029419,
|
3795 |
-
"learning_rate": 2.074468085106383e-07,
|
3796 |
-
"loss": 2.0888,
|
3797 |
-
"step": 540
|
3798 |
-
},
|
3799 |
-
{
|
3800 |
-
"epoch": 1.8161980696600923,
|
3801 |
-
"grad_norm": 0.5545912981033325,
|
3802 |
-
"learning_rate": 2.0685579196217492e-07,
|
3803 |
-
"loss": 2.145,
|
3804 |
-
"step": 541
|
3805 |
-
},
|
3806 |
-
{
|
3807 |
-
"epoch": 1.819555182543013,
|
3808 |
-
"grad_norm": 0.5699747204780579,
|
3809 |
-
"learning_rate": 2.0626477541371157e-07,
|
3810 |
-
"loss": 2.1653,
|
3811 |
-
"step": 542
|
3812 |
-
},
|
3813 |
-
{
|
3814 |
-
"epoch": 1.8229122954259336,
|
3815 |
-
"grad_norm": 0.5698461532592773,
|
3816 |
-
"learning_rate": 2.0567375886524822e-07,
|
3817 |
-
"loss": 2.1712,
|
3818 |
-
"step": 543
|
3819 |
-
},
|
3820 |
-
{
|
3821 |
-
"epoch": 1.8262694083088544,
|
3822 |
-
"grad_norm": 0.5370389223098755,
|
3823 |
-
"learning_rate": 2.0508274231678484e-07,
|
3824 |
-
"loss": 2.1847,
|
3825 |
-
"step": 544
|
3826 |
-
},
|
3827 |
-
{
|
3828 |
-
"epoch": 1.829626521191775,
|
3829 |
-
"grad_norm": 0.5606624484062195,
|
3830 |
-
"learning_rate": 2.044917257683215e-07,
|
3831 |
-
"loss": 2.0752,
|
3832 |
-
"step": 545
|
3833 |
-
},
|
3834 |
-
{
|
3835 |
-
"epoch": 1.8329836340746959,
|
3836 |
-
"grad_norm": 0.5884568691253662,
|
3837 |
-
"learning_rate": 2.0390070921985815e-07,
|
3838 |
-
"loss": 2.1874,
|
3839 |
-
"step": 546
|
3840 |
-
},
|
3841 |
-
{
|
3842 |
-
"epoch": 1.8363407469576165,
|
3843 |
-
"grad_norm": 0.6082828640937805,
|
3844 |
-
"learning_rate": 2.0330969267139477e-07,
|
3845 |
-
"loss": 2.1356,
|
3846 |
-
"step": 547
|
3847 |
-
},
|
3848 |
-
{
|
3849 |
-
"epoch": 1.8396978598405371,
|
3850 |
-
"grad_norm": 0.5440762042999268,
|
3851 |
-
"learning_rate": 2.0271867612293142e-07,
|
3852 |
-
"loss": 2.0869,
|
3853 |
-
"step": 548
|
3854 |
-
},
|
3855 |
-
{
|
3856 |
-
"epoch": 1.8430549727234578,
|
3857 |
-
"grad_norm": 0.5322846174240112,
|
3858 |
-
"learning_rate": 2.0212765957446807e-07,
|
3859 |
-
"loss": 2.1622,
|
3860 |
-
"step": 549
|
3861 |
-
},
|
3862 |
-
{
|
3863 |
-
"epoch": 1.8464120856063784,
|
3864 |
-
"grad_norm": 0.5313152074813843,
|
3865 |
-
"learning_rate": 2.0153664302600472e-07,
|
3866 |
-
"loss": 2.0519,
|
3867 |
-
"step": 550
|
3868 |
-
},
|
3869 |
-
{
|
3870 |
-
"epoch": 1.8497691984892992,
|
3871 |
-
"grad_norm": 0.5677706599235535,
|
3872 |
-
"learning_rate": 2.0094562647754137e-07,
|
3873 |
-
"loss": 2.2212,
|
3874 |
-
"step": 551
|
3875 |
-
},
|
3876 |
-
{
|
3877 |
-
"epoch": 1.8531263113722198,
|
3878 |
-
"grad_norm": 0.5602107644081116,
|
3879 |
-
"learning_rate": 2.0035460992907802e-07,
|
3880 |
-
"loss": 2.1296,
|
3881 |
-
"step": 552
|
3882 |
-
},
|
3883 |
-
{
|
3884 |
-
"epoch": 1.8564834242551407,
|
3885 |
-
"grad_norm": 0.5556351542472839,
|
3886 |
-
"learning_rate": 1.9976359338061465e-07,
|
3887 |
-
"loss": 2.1571,
|
3888 |
-
"step": 553
|
3889 |
-
},
|
3890 |
-
{
|
3891 |
-
"epoch": 1.8598405371380613,
|
3892 |
-
"grad_norm": 0.5427154898643494,
|
3893 |
-
"learning_rate": 1.991725768321513e-07,
|
3894 |
-
"loss": 2.1372,
|
3895 |
-
"step": 554
|
3896 |
-
},
|
3897 |
-
{
|
3898 |
-
"epoch": 1.863197650020982,
|
3899 |
-
"grad_norm": 0.5443682670593262,
|
3900 |
-
"learning_rate": 1.9858156028368795e-07,
|
3901 |
-
"loss": 2.1526,
|
3902 |
-
"step": 555
|
3903 |
-
},
|
3904 |
-
{
|
3905 |
-
"epoch": 1.8665547629039025,
|
3906 |
-
"grad_norm": 0.5522770285606384,
|
3907 |
-
"learning_rate": 1.979905437352246e-07,
|
3908 |
-
"loss": 2.1571,
|
3909 |
-
"step": 556
|
3910 |
-
},
|
3911 |
-
{
|
3912 |
-
"epoch": 1.8699118757868234,
|
3913 |
-
"grad_norm": 0.5588672161102295,
|
3914 |
-
"learning_rate": 1.9739952718676122e-07,
|
3915 |
-
"loss": 2.1527,
|
3916 |
-
"step": 557
|
3917 |
-
},
|
3918 |
-
{
|
3919 |
-
"epoch": 1.873268988669744,
|
3920 |
-
"grad_norm": 0.6060850620269775,
|
3921 |
-
"learning_rate": 1.9680851063829787e-07,
|
3922 |
-
"loss": 2.1447,
|
3923 |
-
"step": 558
|
3924 |
-
},
|
3925 |
-
{
|
3926 |
-
"epoch": 1.8766261015526648,
|
3927 |
-
"grad_norm": 0.5418441295623779,
|
3928 |
-
"learning_rate": 1.9621749408983452e-07,
|
3929 |
-
"loss": 2.1542,
|
3930 |
-
"step": 559
|
3931 |
-
},
|
3932 |
-
{
|
3933 |
-
"epoch": 1.8799832144355855,
|
3934 |
-
"grad_norm": 0.568583607673645,
|
3935 |
-
"learning_rate": 1.9562647754137115e-07,
|
3936 |
-
"loss": 2.2024,
|
3937 |
-
"step": 560
|
3938 |
-
},
|
3939 |
-
{
|
3940 |
-
"epoch": 1.883340327318506,
|
3941 |
-
"grad_norm": 0.566670298576355,
|
3942 |
-
"learning_rate": 1.950354609929078e-07,
|
3943 |
-
"loss": 2.1991,
|
3944 |
-
"step": 561
|
3945 |
-
},
|
3946 |
-
{
|
3947 |
-
"epoch": 1.8866974402014267,
|
3948 |
-
"grad_norm": 0.5521290898323059,
|
3949 |
-
"learning_rate": 1.9444444444444445e-07,
|
3950 |
-
"loss": 2.2061,
|
3951 |
-
"step": 562
|
3952 |
-
},
|
3953 |
-
{
|
3954 |
-
"epoch": 1.8900545530843473,
|
3955 |
-
"grad_norm": 0.5575881600379944,
|
3956 |
-
"learning_rate": 1.9385342789598107e-07,
|
3957 |
-
"loss": 2.1835,
|
3958 |
-
"step": 563
|
3959 |
-
},
|
3960 |
-
{
|
3961 |
-
"epoch": 1.8934116659672682,
|
3962 |
-
"grad_norm": 0.5508525967597961,
|
3963 |
-
"learning_rate": 1.9326241134751772e-07,
|
3964 |
-
"loss": 2.0749,
|
3965 |
-
"step": 564
|
3966 |
-
},
|
3967 |
-
{
|
3968 |
-
"epoch": 1.8967687788501888,
|
3969 |
-
"grad_norm": 0.539969801902771,
|
3970 |
-
"learning_rate": 1.9267139479905437e-07,
|
3971 |
-
"loss": 2.1327,
|
3972 |
-
"step": 565
|
3973 |
-
},
|
3974 |
-
{
|
3975 |
-
"epoch": 1.9001258917331096,
|
3976 |
-
"grad_norm": 0.581095814704895,
|
3977 |
-
"learning_rate": 1.92080378250591e-07,
|
3978 |
-
"loss": 2.097,
|
3979 |
-
"step": 566
|
3980 |
-
},
|
3981 |
-
{
|
3982 |
-
"epoch": 1.9034830046160303,
|
3983 |
-
"grad_norm": 0.5551554560661316,
|
3984 |
-
"learning_rate": 1.9148936170212765e-07,
|
3985 |
-
"loss": 2.2112,
|
3986 |
-
"step": 567
|
3987 |
-
},
|
3988 |
-
{
|
3989 |
-
"epoch": 1.9068401174989509,
|
3990 |
-
"grad_norm": 0.5533036589622498,
|
3991 |
-
"learning_rate": 1.908983451536643e-07,
|
3992 |
-
"loss": 2.1712,
|
3993 |
-
"step": 568
|
3994 |
-
},
|
3995 |
-
{
|
3996 |
-
"epoch": 1.9101972303818715,
|
3997 |
-
"grad_norm": 0.5222402811050415,
|
3998 |
-
"learning_rate": 1.9030732860520092e-07,
|
3999 |
-
"loss": 2.1152,
|
4000 |
-
"step": 569
|
4001 |
-
},
|
4002 |
-
{
|
4003 |
-
"epoch": 1.9135543432647921,
|
4004 |
-
"grad_norm": 0.5835478901863098,
|
4005 |
-
"learning_rate": 1.8971631205673757e-07,
|
4006 |
-
"loss": 2.0906,
|
4007 |
-
"step": 570
|
4008 |
-
},
|
4009 |
-
{
|
4010 |
-
"epoch": 1.916911456147713,
|
4011 |
-
"grad_norm": 0.5896159410476685,
|
4012 |
-
"learning_rate": 1.8912529550827422e-07,
|
4013 |
-
"loss": 2.1444,
|
4014 |
-
"step": 571
|
4015 |
-
},
|
4016 |
-
{
|
4017 |
-
"epoch": 1.9202685690306338,
|
4018 |
-
"grad_norm": 0.5776488184928894,
|
4019 |
-
"learning_rate": 1.8853427895981087e-07,
|
4020 |
-
"loss": 2.163,
|
4021 |
-
"step": 572
|
4022 |
-
},
|
4023 |
-
{
|
4024 |
-
"epoch": 1.9236256819135544,
|
4025 |
-
"grad_norm": 0.6126729249954224,
|
4026 |
-
"learning_rate": 1.879432624113475e-07,
|
4027 |
-
"loss": 2.144,
|
4028 |
-
"step": 573
|
4029 |
-
},
|
4030 |
-
{
|
4031 |
-
"epoch": 1.926982794796475,
|
4032 |
-
"grad_norm": 0.5623863339424133,
|
4033 |
-
"learning_rate": 1.8735224586288415e-07,
|
4034 |
-
"loss": 2.1958,
|
4035 |
-
"step": 574
|
4036 |
-
},
|
4037 |
-
{
|
4038 |
-
"epoch": 1.9303399076793957,
|
4039 |
-
"grad_norm": 0.5552009344100952,
|
4040 |
-
"learning_rate": 1.867612293144208e-07,
|
4041 |
-
"loss": 2.128,
|
4042 |
-
"step": 575
|
4043 |
-
},
|
4044 |
-
{
|
4045 |
-
"epoch": 1.9336970205623163,
|
4046 |
-
"grad_norm": 0.5190218687057495,
|
4047 |
-
"learning_rate": 1.8617021276595742e-07,
|
4048 |
-
"loss": 2.1553,
|
4049 |
-
"step": 576
|
4050 |
-
},
|
4051 |
-
{
|
4052 |
-
"epoch": 1.9370541334452371,
|
4053 |
-
"grad_norm": 0.5255279541015625,
|
4054 |
-
"learning_rate": 1.8557919621749407e-07,
|
4055 |
-
"loss": 2.148,
|
4056 |
-
"step": 577
|
4057 |
-
},
|
4058 |
-
{
|
4059 |
-
"epoch": 1.9404112463281578,
|
4060 |
-
"grad_norm": 0.5752416253089905,
|
4061 |
-
"learning_rate": 1.8498817966903072e-07,
|
4062 |
-
"loss": 2.1934,
|
4063 |
-
"step": 578
|
4064 |
-
},
|
4065 |
-
{
|
4066 |
-
"epoch": 1.9437683592110786,
|
4067 |
-
"grad_norm": 0.532777726650238,
|
4068 |
-
"learning_rate": 1.8439716312056735e-07,
|
4069 |
-
"loss": 2.0636,
|
4070 |
-
"step": 579
|
4071 |
-
},
|
4072 |
-
{
|
4073 |
-
"epoch": 1.9471254720939992,
|
4074 |
-
"grad_norm": 0.5603284239768982,
|
4075 |
-
"learning_rate": 1.83806146572104e-07,
|
4076 |
-
"loss": 2.1411,
|
4077 |
-
"step": 580
|
4078 |
-
},
|
4079 |
-
{
|
4080 |
-
"epoch": 1.9504825849769198,
|
4081 |
-
"grad_norm": 0.5280658006668091,
|
4082 |
-
"learning_rate": 1.8321513002364065e-07,
|
4083 |
-
"loss": 2.1211,
|
4084 |
-
"step": 581
|
4085 |
-
},
|
4086 |
-
{
|
4087 |
-
"epoch": 1.9538396978598405,
|
4088 |
-
"grad_norm": 0.5879082679748535,
|
4089 |
-
"learning_rate": 1.826241134751773e-07,
|
4090 |
-
"loss": 2.1186,
|
4091 |
-
"step": 582
|
4092 |
-
},
|
4093 |
-
{
|
4094 |
-
"epoch": 1.957196810742761,
|
4095 |
-
"grad_norm": 0.5637861490249634,
|
4096 |
-
"learning_rate": 1.8203309692671395e-07,
|
4097 |
-
"loss": 2.112,
|
4098 |
-
"step": 583
|
4099 |
-
},
|
4100 |
-
{
|
4101 |
-
"epoch": 1.960553923625682,
|
4102 |
-
"grad_norm": 0.5610608458518982,
|
4103 |
-
"learning_rate": 1.814420803782506e-07,
|
4104 |
-
"loss": 2.1645,
|
4105 |
-
"step": 584
|
4106 |
-
},
|
4107 |
-
{
|
4108 |
-
"epoch": 1.9639110365086025,
|
4109 |
-
"grad_norm": 0.5636882185935974,
|
4110 |
-
"learning_rate": 1.8085106382978725e-07,
|
4111 |
-
"loss": 2.1192,
|
4112 |
-
"step": 585
|
4113 |
-
},
|
4114 |
-
{
|
4115 |
-
"epoch": 1.9672681493915234,
|
4116 |
-
"grad_norm": 0.5535016059875488,
|
4117 |
-
"learning_rate": 1.8026004728132387e-07,
|
4118 |
-
"loss": 2.1384,
|
4119 |
-
"step": 586
|
4120 |
-
},
|
4121 |
-
{
|
4122 |
-
"epoch": 1.970625262274444,
|
4123 |
-
"grad_norm": 0.5968726873397827,
|
4124 |
-
"learning_rate": 1.7966903073286052e-07,
|
4125 |
-
"loss": 2.1896,
|
4126 |
-
"step": 587
|
4127 |
-
},
|
4128 |
-
{
|
4129 |
-
"epoch": 1.9739823751573646,
|
4130 |
-
"grad_norm": 0.5755752325057983,
|
4131 |
-
"learning_rate": 1.7907801418439718e-07,
|
4132 |
-
"loss": 2.1582,
|
4133 |
-
"step": 588
|
4134 |
-
},
|
4135 |
-
{
|
4136 |
-
"epoch": 1.9773394880402853,
|
4137 |
-
"grad_norm": 0.5775538086891174,
|
4138 |
-
"learning_rate": 1.784869976359338e-07,
|
4139 |
-
"loss": 2.1839,
|
4140 |
-
"step": 589
|
4141 |
-
},
|
4142 |
-
{
|
4143 |
-
"epoch": 1.980696600923206,
|
4144 |
-
"grad_norm": 0.5270890593528748,
|
4145 |
-
"learning_rate": 1.7789598108747045e-07,
|
4146 |
-
"loss": 2.083,
|
4147 |
-
"step": 590
|
4148 |
-
},
|
4149 |
-
{
|
4150 |
-
"epoch": 1.9840537138061267,
|
4151 |
-
"grad_norm": 0.5608636140823364,
|
4152 |
-
"learning_rate": 1.773049645390071e-07,
|
4153 |
-
"loss": 2.1341,
|
4154 |
-
"step": 591
|
4155 |
-
},
|
4156 |
-
{
|
4157 |
-
"epoch": 1.9874108266890476,
|
4158 |
-
"grad_norm": 0.5662211179733276,
|
4159 |
-
"learning_rate": 1.7671394799054372e-07,
|
4160 |
-
"loss": 2.151,
|
4161 |
-
"step": 592
|
4162 |
-
},
|
4163 |
-
{
|
4164 |
-
"epoch": 1.9907679395719682,
|
4165 |
-
"grad_norm": 0.5171293020248413,
|
4166 |
-
"learning_rate": 1.7612293144208037e-07,
|
4167 |
-
"loss": 2.189,
|
4168 |
-
"step": 593
|
4169 |
-
},
|
4170 |
-
{
|
4171 |
-
"epoch": 1.9941250524548888,
|
4172 |
-
"grad_norm": 0.543449878692627,
|
4173 |
-
"learning_rate": 1.7553191489361703e-07,
|
4174 |
-
"loss": 2.0942,
|
4175 |
-
"step": 594
|
4176 |
-
},
|
4177 |
-
{
|
4178 |
-
"epoch": 1.9974821653378094,
|
4179 |
-
"grad_norm": 0.5563431978225708,
|
4180 |
-
"learning_rate": 1.7494089834515365e-07,
|
4181 |
-
"loss": 2.1623,
|
4182 |
-
"step": 595
|
4183 |
-
},
|
4184 |
-
{
|
4185 |
-
"epoch": 1.9974821653378094,
|
4186 |
-
"eval_loss": 2.154229164123535,
|
4187 |
-
"eval_runtime": 325.0894,
|
4188 |
-
"eval_samples_per_second": 1.114,
|
4189 |
-
"eval_steps_per_second": 0.28,
|
4190 |
-
"step": 595
|
4191 |
}
|
4192 |
],
|
4193 |
"logging_steps": 1,
|
@@ -4207,7 +2113,7 @@
|
|
4207 |
"attributes": {}
|
4208 |
}
|
4209 |
},
|
4210 |
-
"total_flos":
|
4211 |
"train_batch_size": 2,
|
4212 |
"trial_name": null,
|
4213 |
"trial_params": null
|
|
|
1 |
{
|
2 |
+
"best_metric": 2.169156074523926,
|
3 |
+
"best_model_checkpoint": "/home/sunggeunan/data/ICL/outputs/lora/SKIML-ICL_mrqa_nq_v3/Meta-Llama-3-8B-Instruct-unanswerable-1Q-0U-0C-qa_first/checkpoint-297",
|
4 |
+
"epoch": 0.9970625262274444,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 297,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
2094 |
"eval_samples_per_second": 1.004,
|
2095 |
"eval_steps_per_second": 0.252,
|
2096 |
"step": 297
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2097 |
}
|
2098 |
],
|
2099 |
"logging_steps": 1,
|
|
|
2113 |
"attributes": {}
|
2114 |
}
|
2115 |
},
|
2116 |
+
"total_flos": 8.794451043460055e+17,
|
2117 |
"train_batch_size": 2,
|
2118 |
"trial_name": null,
|
2119 |
"trial_params": null
|