Training in progress, step 36000, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3555504
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4a4b1c16a17272dd3d5d44145b877d0c14dc6e19c78bd41637a76bc87a0e7ea
|
| 3 |
size 3555504
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7141515
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68ca690b9a0c619477d3038e9b1f2189091b893c636b800c671b19aeb6d28dba
|
| 3 |
size 7141515
|
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 sha256:4d04e325551f27b0abb1677ad51b5250cc6db4cb37a98cc2cc710661788306e4
|
| 3 |
size 14645
|
last-checkpoint/scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1383
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eba75c9103403d8e6ccb33e21824c3bb41c5512582b8b3e0415b8a731bc74943
|
| 3 |
size 1383
|
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 sha256:4bef14adf96a5be9d64a1df76f457ac5b2fcd08c7a9b48176945077bc67a3f2f
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 0.
|
| 4 |
-
"best_model_checkpoint": "CodeGenDetect-CodeBert_Lora/checkpoint-
|
| 5 |
-
"epoch": 1.
|
| 6 |
"eval_steps": 4000,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -2344,6 +2344,298 @@
|
|
| 2344 |
"eval_samples_per_second": 128.957,
|
| 2345 |
"eval_steps_per_second": 8.06,
|
| 2346 |
"step": 32000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2347 |
}
|
| 2348 |
],
|
| 2349 |
"logging_steps": 100,
|
|
@@ -2372,7 +2664,7 @@
|
|
| 2372 |
"attributes": {}
|
| 2373 |
}
|
| 2374 |
},
|
| 2375 |
-
"total_flos": 1.
|
| 2376 |
"train_batch_size": 16,
|
| 2377 |
"trial_name": null,
|
| 2378 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 36000,
|
| 3 |
+
"best_metric": 0.9893807849919393,
|
| 4 |
+
"best_model_checkpoint": "CodeGenDetect-CodeBert_Lora/checkpoint-36000",
|
| 5 |
+
"epoch": 1.152,
|
| 6 |
"eval_steps": 4000,
|
| 7 |
+
"global_step": 36000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 2344 |
"eval_samples_per_second": 128.957,
|
| 2345 |
"eval_steps_per_second": 8.06,
|
| 2346 |
"step": 32000
|
| 2347 |
+
},
|
| 2348 |
+
{
|
| 2349 |
+
"epoch": 1.0272,
|
| 2350 |
+
"grad_norm": 1.0557373762130737,
|
| 2351 |
+
"learning_rate": 1.5942343499197433e-05,
|
| 2352 |
+
"loss": 0.0594,
|
| 2353 |
+
"step": 32100
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 1.0304,
|
| 2357 |
+
"grad_norm": 0.03562343493103981,
|
| 2358 |
+
"learning_rate": 1.5929502407704655e-05,
|
| 2359 |
+
"loss": 0.0537,
|
| 2360 |
+
"step": 32200
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 1.0336,
|
| 2364 |
+
"grad_norm": 0.06482692807912827,
|
| 2365 |
+
"learning_rate": 1.591666131621188e-05,
|
| 2366 |
+
"loss": 0.0463,
|
| 2367 |
+
"step": 32300
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 1.0368,
|
| 2371 |
+
"grad_norm": 0.49017927050590515,
|
| 2372 |
+
"learning_rate": 1.59038202247191e-05,
|
| 2373 |
+
"loss": 0.0661,
|
| 2374 |
+
"step": 32400
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 1.04,
|
| 2378 |
+
"grad_norm": 15.849417686462402,
|
| 2379 |
+
"learning_rate": 1.5890979133226326e-05,
|
| 2380 |
+
"loss": 0.0237,
|
| 2381 |
+
"step": 32500
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 1.0432,
|
| 2385 |
+
"grad_norm": 0.6910932064056396,
|
| 2386 |
+
"learning_rate": 1.5878138041733547e-05,
|
| 2387 |
+
"loss": 0.0383,
|
| 2388 |
+
"step": 32600
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 1.0464,
|
| 2392 |
+
"grad_norm": 0.06304027885198593,
|
| 2393 |
+
"learning_rate": 1.5865296950240772e-05,
|
| 2394 |
+
"loss": 0.0541,
|
| 2395 |
+
"step": 32700
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 1.0496,
|
| 2399 |
+
"grad_norm": 11.758397102355957,
|
| 2400 |
+
"learning_rate": 1.5852455858747994e-05,
|
| 2401 |
+
"loss": 0.0573,
|
| 2402 |
+
"step": 32800
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 1.0528,
|
| 2406 |
+
"grad_norm": 0.007580827921628952,
|
| 2407 |
+
"learning_rate": 1.583961476725522e-05,
|
| 2408 |
+
"loss": 0.0551,
|
| 2409 |
+
"step": 32900
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 1.056,
|
| 2413 |
+
"grad_norm": 0.7389895915985107,
|
| 2414 |
+
"learning_rate": 1.582677367576244e-05,
|
| 2415 |
+
"loss": 0.0408,
|
| 2416 |
+
"step": 33000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 1.0592,
|
| 2420 |
+
"grad_norm": 0.011490071192383766,
|
| 2421 |
+
"learning_rate": 1.5813932584269665e-05,
|
| 2422 |
+
"loss": 0.0474,
|
| 2423 |
+
"step": 33100
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 1.0624,
|
| 2427 |
+
"grad_norm": 0.00892989058047533,
|
| 2428 |
+
"learning_rate": 1.5801091492776886e-05,
|
| 2429 |
+
"loss": 0.0338,
|
| 2430 |
+
"step": 33200
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 1.0656,
|
| 2434 |
+
"grad_norm": 0.21890480816364288,
|
| 2435 |
+
"learning_rate": 1.578825040128411e-05,
|
| 2436 |
+
"loss": 0.0465,
|
| 2437 |
+
"step": 33300
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 1.0688,
|
| 2441 |
+
"grad_norm": 20.05085563659668,
|
| 2442 |
+
"learning_rate": 1.5775409309791332e-05,
|
| 2443 |
+
"loss": 0.0633,
|
| 2444 |
+
"step": 33400
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 1.072,
|
| 2448 |
+
"grad_norm": 3.5141100883483887,
|
| 2449 |
+
"learning_rate": 1.5762568218298557e-05,
|
| 2450 |
+
"loss": 0.0743,
|
| 2451 |
+
"step": 33500
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 1.0752,
|
| 2455 |
+
"grad_norm": 0.03382471203804016,
|
| 2456 |
+
"learning_rate": 1.5749727126805782e-05,
|
| 2457 |
+
"loss": 0.0409,
|
| 2458 |
+
"step": 33600
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 1.0784,
|
| 2462 |
+
"grad_norm": 9.916868209838867,
|
| 2463 |
+
"learning_rate": 1.5736886035313004e-05,
|
| 2464 |
+
"loss": 0.0498,
|
| 2465 |
+
"step": 33700
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 1.0816,
|
| 2469 |
+
"grad_norm": 0.007973396219313145,
|
| 2470 |
+
"learning_rate": 1.572404494382023e-05,
|
| 2471 |
+
"loss": 0.0399,
|
| 2472 |
+
"step": 33800
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 1.0848,
|
| 2476 |
+
"grad_norm": 0.15778931975364685,
|
| 2477 |
+
"learning_rate": 1.571120385232745e-05,
|
| 2478 |
+
"loss": 0.0406,
|
| 2479 |
+
"step": 33900
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 1.088,
|
| 2483 |
+
"grad_norm": 0.027699623256921768,
|
| 2484 |
+
"learning_rate": 1.569836276083467e-05,
|
| 2485 |
+
"loss": 0.0481,
|
| 2486 |
+
"step": 34000
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 1.0912,
|
| 2490 |
+
"grad_norm": 2.7082841396331787,
|
| 2491 |
+
"learning_rate": 1.5685521669341893e-05,
|
| 2492 |
+
"loss": 0.0427,
|
| 2493 |
+
"step": 34100
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 1.0944,
|
| 2497 |
+
"grad_norm": 2.933098316192627,
|
| 2498 |
+
"learning_rate": 1.5672680577849118e-05,
|
| 2499 |
+
"loss": 0.0605,
|
| 2500 |
+
"step": 34200
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 1.0976,
|
| 2504 |
+
"grad_norm": 5.093338489532471,
|
| 2505 |
+
"learning_rate": 1.5659839486356342e-05,
|
| 2506 |
+
"loss": 0.0625,
|
| 2507 |
+
"step": 34300
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 1.1008,
|
| 2511 |
+
"grad_norm": 0.14764878153800964,
|
| 2512 |
+
"learning_rate": 1.5646998394863564e-05,
|
| 2513 |
+
"loss": 0.0467,
|
| 2514 |
+
"step": 34400
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 1.104,
|
| 2518 |
+
"grad_norm": 0.45360442996025085,
|
| 2519 |
+
"learning_rate": 1.563415730337079e-05,
|
| 2520 |
+
"loss": 0.0517,
|
| 2521 |
+
"step": 34500
|
| 2522 |
+
},
|
| 2523 |
+
{
|
| 2524 |
+
"epoch": 1.1072,
|
| 2525 |
+
"grad_norm": 0.007702784612774849,
|
| 2526 |
+
"learning_rate": 1.562131621187801e-05,
|
| 2527 |
+
"loss": 0.0367,
|
| 2528 |
+
"step": 34600
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 1.1104,
|
| 2532 |
+
"grad_norm": 0.01513522956520319,
|
| 2533 |
+
"learning_rate": 1.5608475120385235e-05,
|
| 2534 |
+
"loss": 0.0514,
|
| 2535 |
+
"step": 34700
|
| 2536 |
+
},
|
| 2537 |
+
{
|
| 2538 |
+
"epoch": 1.1136,
|
| 2539 |
+
"grad_norm": 0.29241982102394104,
|
| 2540 |
+
"learning_rate": 1.5595634028892456e-05,
|
| 2541 |
+
"loss": 0.0428,
|
| 2542 |
+
"step": 34800
|
| 2543 |
+
},
|
| 2544 |
+
{
|
| 2545 |
+
"epoch": 1.1168,
|
| 2546 |
+
"grad_norm": 0.02503369376063347,
|
| 2547 |
+
"learning_rate": 1.558279293739968e-05,
|
| 2548 |
+
"loss": 0.0583,
|
| 2549 |
+
"step": 34900
|
| 2550 |
+
},
|
| 2551 |
+
{
|
| 2552 |
+
"epoch": 1.12,
|
| 2553 |
+
"grad_norm": 0.094393789768219,
|
| 2554 |
+
"learning_rate": 1.5569951845906903e-05,
|
| 2555 |
+
"loss": 0.0636,
|
| 2556 |
+
"step": 35000
|
| 2557 |
+
},
|
| 2558 |
+
{
|
| 2559 |
+
"epoch": 1.1232,
|
| 2560 |
+
"grad_norm": 0.011400578543543816,
|
| 2561 |
+
"learning_rate": 1.5557110754414127e-05,
|
| 2562 |
+
"loss": 0.0392,
|
| 2563 |
+
"step": 35100
|
| 2564 |
+
},
|
| 2565 |
+
{
|
| 2566 |
+
"epoch": 1.1264,
|
| 2567 |
+
"grad_norm": 0.015260876156389713,
|
| 2568 |
+
"learning_rate": 1.554426966292135e-05,
|
| 2569 |
+
"loss": 0.0452,
|
| 2570 |
+
"step": 35200
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 1.1296,
|
| 2574 |
+
"grad_norm": 0.07425595074892044,
|
| 2575 |
+
"learning_rate": 1.5531428571428574e-05,
|
| 2576 |
+
"loss": 0.0506,
|
| 2577 |
+
"step": 35300
|
| 2578 |
+
},
|
| 2579 |
+
{
|
| 2580 |
+
"epoch": 1.1328,
|
| 2581 |
+
"grad_norm": 2.20249605178833,
|
| 2582 |
+
"learning_rate": 1.5518587479935795e-05,
|
| 2583 |
+
"loss": 0.0742,
|
| 2584 |
+
"step": 35400
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 1.1360000000000001,
|
| 2588 |
+
"grad_norm": 0.027821656316518784,
|
| 2589 |
+
"learning_rate": 1.550574638844302e-05,
|
| 2590 |
+
"loss": 0.062,
|
| 2591 |
+
"step": 35500
|
| 2592 |
+
},
|
| 2593 |
+
{
|
| 2594 |
+
"epoch": 1.1392,
|
| 2595 |
+
"grad_norm": 0.02665848284959793,
|
| 2596 |
+
"learning_rate": 1.549290529695024e-05,
|
| 2597 |
+
"loss": 0.0739,
|
| 2598 |
+
"step": 35600
|
| 2599 |
+
},
|
| 2600 |
+
{
|
| 2601 |
+
"epoch": 1.1424,
|
| 2602 |
+
"grad_norm": 5.973049163818359,
|
| 2603 |
+
"learning_rate": 1.5480064205457466e-05,
|
| 2604 |
+
"loss": 0.0504,
|
| 2605 |
+
"step": 35700
|
| 2606 |
+
},
|
| 2607 |
+
{
|
| 2608 |
+
"epoch": 1.1456,
|
| 2609 |
+
"grad_norm": 10.988951683044434,
|
| 2610 |
+
"learning_rate": 1.5467223113964688e-05,
|
| 2611 |
+
"loss": 0.0493,
|
| 2612 |
+
"step": 35800
|
| 2613 |
+
},
|
| 2614 |
+
{
|
| 2615 |
+
"epoch": 1.1488,
|
| 2616 |
+
"grad_norm": 15.630033493041992,
|
| 2617 |
+
"learning_rate": 1.5454382022471913e-05,
|
| 2618 |
+
"loss": 0.0492,
|
| 2619 |
+
"step": 35900
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 1.152,
|
| 2623 |
+
"grad_norm": 4.891533851623535,
|
| 2624 |
+
"learning_rate": 1.5441540930979134e-05,
|
| 2625 |
+
"loss": 0.0427,
|
| 2626 |
+
"step": 36000
|
| 2627 |
+
},
|
| 2628 |
+
{
|
| 2629 |
+
"epoch": 1.152,
|
| 2630 |
+
"eval_accuracy": 0.98938,
|
| 2631 |
+
"eval_f1": 0.9893807849919393,
|
| 2632 |
+
"eval_loss": 0.0424417182803154,
|
| 2633 |
+
"eval_precision": 0.9893870324245919,
|
| 2634 |
+
"eval_recall": 0.98938,
|
| 2635 |
+
"eval_runtime": 778.7856,
|
| 2636 |
+
"eval_samples_per_second": 128.405,
|
| 2637 |
+
"eval_steps_per_second": 8.025,
|
| 2638 |
+
"step": 36000
|
| 2639 |
}
|
| 2640 |
],
|
| 2641 |
"logging_steps": 100,
|
|
|
|
| 2664 |
"attributes": {}
|
| 2665 |
}
|
| 2666 |
},
|
| 2667 |
+
"total_flos": 1.528656883652379e+17,
|
| 2668 |
"train_batch_size": 16,
|
| 2669 |
"trial_name": null,
|
| 2670 |
"trial_params": null
|