Training in progress, step 18000, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1520630616
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0cc7670971a4e72c8112661889eb09f4dd0aef80b62662f3de10aa7539b0126d
|
| 3 |
size 1520630616
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3041448587
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7fa9da6182eb634e04ef5b1e88adde1464e5173eb3d0eb66d882cdcb5bad981e
|
| 3 |
size 3041448587
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:59533f6351ee8031370e4d884b963a0d20ec1f96789095c1a70c9891c4bf2301
|
| 3 |
size 14645
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1fcb8a6684d40f0d667cf5f41391378b3c7ac8a01224006fa696bc503a2d8b3b
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -2248,6 +2248,286 @@
|
|
| 2248 |
"learning_rate": 0.0002830845786762962,
|
| 2249 |
"loss": 1.7125,
|
| 2250 |
"step": 16000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2251 |
}
|
| 2252 |
],
|
| 2253 |
"logging_steps": 50,
|
|
@@ -2267,7 +2547,7 @@
|
|
| 2267 |
"attributes": {}
|
| 2268 |
}
|
| 2269 |
},
|
| 2270 |
-
"total_flos":
|
| 2271 |
"train_batch_size": 16,
|
| 2272 |
"trial_name": null,
|
| 2273 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 5.008579234972678,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 18000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 2248 |
"learning_rate": 0.0002830845786762962,
|
| 2249 |
"loss": 1.7125,
|
| 2250 |
"step": 16000
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 4.024896174863388,
|
| 2254 |
+
"grad_norm": 0.6796875,
|
| 2255 |
+
"learning_rate": 0.0002829623770208463,
|
| 2256 |
+
"loss": 1.6678,
|
| 2257 |
+
"step": 16050
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 4.02544262295082,
|
| 2261 |
+
"grad_norm": 0.55859375,
|
| 2262 |
+
"learning_rate": 0.00028283976211526137,
|
| 2263 |
+
"loss": 1.7396,
|
| 2264 |
+
"step": 16100
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 4.025989071038252,
|
| 2268 |
+
"grad_norm": 0.578125,
|
| 2269 |
+
"learning_rate": 0.0002827167343406315,
|
| 2270 |
+
"loss": 1.752,
|
| 2271 |
+
"step": 16150
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 4.026535519125683,
|
| 2275 |
+
"grad_norm": 0.56640625,
|
| 2276 |
+
"learning_rate": 0.0002825932940793298,
|
| 2277 |
+
"loss": 1.6994,
|
| 2278 |
+
"step": 16200
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 4.027081967213115,
|
| 2282 |
+
"grad_norm": 0.6015625,
|
| 2283 |
+
"learning_rate": 0.00028246944171501145,
|
| 2284 |
+
"loss": 1.7189,
|
| 2285 |
+
"step": 16250
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 4.027628415300547,
|
| 2289 |
+
"grad_norm": 0.5625,
|
| 2290 |
+
"learning_rate": 0.00028234517763261243,
|
| 2291 |
+
"loss": 1.7158,
|
| 2292 |
+
"step": 16300
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 4.028174863387978,
|
| 2296 |
+
"grad_norm": 0.5625,
|
| 2297 |
+
"learning_rate": 0.00028222050221834847,
|
| 2298 |
+
"loss": 1.6885,
|
| 2299 |
+
"step": 16350
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 4.02872131147541,
|
| 2303 |
+
"grad_norm": 0.5078125,
|
| 2304 |
+
"learning_rate": 0.0002820954158597134,
|
| 2305 |
+
"loss": 1.6865,
|
| 2306 |
+
"step": 16400
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 4.029267759562842,
|
| 2310 |
+
"grad_norm": 0.52734375,
|
| 2311 |
+
"learning_rate": 0.0002819699189454788,
|
| 2312 |
+
"loss": 1.6946,
|
| 2313 |
+
"step": 16450
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 4.029814207650273,
|
| 2317 |
+
"grad_norm": 0.5703125,
|
| 2318 |
+
"learning_rate": 0.0002818440118656918,
|
| 2319 |
+
"loss": 1.7194,
|
| 2320 |
+
"step": 16500
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 4.030360655737705,
|
| 2324 |
+
"grad_norm": 0.52734375,
|
| 2325 |
+
"learning_rate": 0.0002817176950116746,
|
| 2326 |
+
"loss": 1.7022,
|
| 2327 |
+
"step": 16550
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 4.030907103825137,
|
| 2331 |
+
"grad_norm": 0.61328125,
|
| 2332 |
+
"learning_rate": 0.00028159096877602275,
|
| 2333 |
+
"loss": 1.7159,
|
| 2334 |
+
"step": 16600
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 4.031453551912568,
|
| 2338 |
+
"grad_norm": 0.5625,
|
| 2339 |
+
"learning_rate": 0.00028146383355260446,
|
| 2340 |
+
"loss": 1.7063,
|
| 2341 |
+
"step": 16650
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 4.032,
|
| 2345 |
+
"grad_norm": 0.58984375,
|
| 2346 |
+
"learning_rate": 0.00028133628973655894,
|
| 2347 |
+
"loss": 1.7064,
|
| 2348 |
+
"step": 16700
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 4.032546448087432,
|
| 2352 |
+
"grad_norm": 0.55859375,
|
| 2353 |
+
"learning_rate": 0.00028120833772429517,
|
| 2354 |
+
"loss": 1.6992,
|
| 2355 |
+
"step": 16750
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 4.033092896174863,
|
| 2359 |
+
"grad_norm": 0.5546875,
|
| 2360 |
+
"learning_rate": 0.0002810799779134911,
|
| 2361 |
+
"loss": 1.7132,
|
| 2362 |
+
"step": 16800
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 4.033639344262295,
|
| 2366 |
+
"grad_norm": 0.5390625,
|
| 2367 |
+
"learning_rate": 0.0002809512107030919,
|
| 2368 |
+
"loss": 1.7125,
|
| 2369 |
+
"step": 16850
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 4.034185792349727,
|
| 2373 |
+
"grad_norm": 0.57421875,
|
| 2374 |
+
"learning_rate": 0.0002808220364933091,
|
| 2375 |
+
"loss": 1.6373,
|
| 2376 |
+
"step": 16900
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 4.034732240437158,
|
| 2380 |
+
"grad_norm": 0.5234375,
|
| 2381 |
+
"learning_rate": 0.00028069245568561904,
|
| 2382 |
+
"loss": 1.7379,
|
| 2383 |
+
"step": 16950
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 4.03527868852459,
|
| 2387 |
+
"grad_norm": 0.6015625,
|
| 2388 |
+
"learning_rate": 0.00028056246868276186,
|
| 2389 |
+
"loss": 1.699,
|
| 2390 |
+
"step": 17000
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 4.035825136612022,
|
| 2394 |
+
"grad_norm": 0.5546875,
|
| 2395 |
+
"learning_rate": 0.0002804320758887403,
|
| 2396 |
+
"loss": 1.6939,
|
| 2397 |
+
"step": 17050
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 4.036371584699453,
|
| 2401 |
+
"grad_norm": 0.498046875,
|
| 2402 |
+
"learning_rate": 0.000280301277708818,
|
| 2403 |
+
"loss": 1.7146,
|
| 2404 |
+
"step": 17100
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 4.036918032786885,
|
| 2408 |
+
"grad_norm": 0.5859375,
|
| 2409 |
+
"learning_rate": 0.00028017007454951884,
|
| 2410 |
+
"loss": 1.7363,
|
| 2411 |
+
"step": 17150
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 4.037464480874317,
|
| 2415 |
+
"grad_norm": 0.55078125,
|
| 2416 |
+
"learning_rate": 0.00028003846681862524,
|
| 2417 |
+
"loss": 1.6965,
|
| 2418 |
+
"step": 17200
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 5.000382513661203,
|
| 2422 |
+
"grad_norm": 0.5703125,
|
| 2423 |
+
"learning_rate": 0.00027990645492517697,
|
| 2424 |
+
"loss": 1.6887,
|
| 2425 |
+
"step": 17250
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 5.000928961748634,
|
| 2429 |
+
"grad_norm": 0.62109375,
|
| 2430 |
+
"learning_rate": 0.0002797740392794702,
|
| 2431 |
+
"loss": 1.6829,
|
| 2432 |
+
"step": 17300
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 5.001475409836066,
|
| 2436 |
+
"grad_norm": 0.58984375,
|
| 2437 |
+
"learning_rate": 0.0002796412202930557,
|
| 2438 |
+
"loss": 1.6235,
|
| 2439 |
+
"step": 17350
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 5.002021857923498,
|
| 2443 |
+
"grad_norm": 0.59375,
|
| 2444 |
+
"learning_rate": 0.00027950799837873794,
|
| 2445 |
+
"loss": 1.634,
|
| 2446 |
+
"step": 17400
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 5.002568306010929,
|
| 2450 |
+
"grad_norm": 0.55859375,
|
| 2451 |
+
"learning_rate": 0.0002793743739505738,
|
| 2452 |
+
"loss": 1.5863,
|
| 2453 |
+
"step": 17450
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 5.003114754098361,
|
| 2457 |
+
"grad_norm": 0.5546875,
|
| 2458 |
+
"learning_rate": 0.0002792403474238709,
|
| 2459 |
+
"loss": 1.6513,
|
| 2460 |
+
"step": 17500
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 5.003661202185793,
|
| 2464 |
+
"grad_norm": 0.59375,
|
| 2465 |
+
"learning_rate": 0.000279105919215187,
|
| 2466 |
+
"loss": 1.6388,
|
| 2467 |
+
"step": 17550
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 5.004207650273224,
|
| 2471 |
+
"grad_norm": 0.59375,
|
| 2472 |
+
"learning_rate": 0.00027897108974232797,
|
| 2473 |
+
"loss": 1.6302,
|
| 2474 |
+
"step": 17600
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 5.004754098360656,
|
| 2478 |
+
"grad_norm": 1.1875,
|
| 2479 |
+
"learning_rate": 0.0002788358594243469,
|
| 2480 |
+
"loss": 1.6336,
|
| 2481 |
+
"step": 17650
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 5.005300546448088,
|
| 2485 |
+
"grad_norm": 0.58984375,
|
| 2486 |
+
"learning_rate": 0.0002787002286815428,
|
| 2487 |
+
"loss": 1.6513,
|
| 2488 |
+
"step": 17700
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 5.005846994535519,
|
| 2492 |
+
"grad_norm": 0.6015625,
|
| 2493 |
+
"learning_rate": 0.000278564197935459,
|
| 2494 |
+
"loss": 1.6185,
|
| 2495 |
+
"step": 17750
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 5.006393442622951,
|
| 2499 |
+
"grad_norm": 0.58984375,
|
| 2500 |
+
"learning_rate": 0.00027842776760888236,
|
| 2501 |
+
"loss": 1.6366,
|
| 2502 |
+
"step": 17800
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 5.006939890710383,
|
| 2506 |
+
"grad_norm": 0.578125,
|
| 2507 |
+
"learning_rate": 0.00027829093812584143,
|
| 2508 |
+
"loss": 1.6162,
|
| 2509 |
+
"step": 17850
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 5.007486338797814,
|
| 2513 |
+
"grad_norm": 0.56640625,
|
| 2514 |
+
"learning_rate": 0.0002781537099116054,
|
| 2515 |
+
"loss": 1.6665,
|
| 2516 |
+
"step": 17900
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 5.008032786885246,
|
| 2520 |
+
"grad_norm": 0.55859375,
|
| 2521 |
+
"learning_rate": 0.00027801608339268275,
|
| 2522 |
+
"loss": 1.6505,
|
| 2523 |
+
"step": 17950
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 5.008579234972678,
|
| 2527 |
+
"grad_norm": 0.578125,
|
| 2528 |
+
"learning_rate": 0.00027787805899681976,
|
| 2529 |
+
"loss": 1.6079,
|
| 2530 |
+
"step": 18000
|
| 2531 |
}
|
| 2532 |
],
|
| 2533 |
"logging_steps": 50,
|
|
|
|
| 2547 |
"attributes": {}
|
| 2548 |
}
|
| 2549 |
},
|
| 2550 |
+
"total_flos": 9.6261070880401e+18,
|
| 2551 |
"train_batch_size": 16,
|
| 2552 |
"trial_name": null,
|
| 2553 |
"trial_params": null
|