Training in progress, step 3160

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:52fb5f5074bb04352a2040e84ac0619fff64c08e3e204a124fd1aaecca299203
+oid sha256:5035ab920a7272b44f46d9988bd81a38546c99f67449ffe36f2cdaef8b259e7d
 size 1638528

trainer_log.jsonl

@@ -620,3 +620,34 @@
 {"current_steps": 3005, "total_steps": 3160, "loss": 0.0005, "lr": 7.405506829382736e-06, "epoch": 19.018987341772153, "percentage": 95.09, "elapsed_time": "0:10:26", "remaining_time": "0:00:32", "throughput": 2170.87, "total_tokens": 1360464}
 {"current_steps": 3010, "total_steps": 3160, "loss": 0.0007, "lr": 6.9394860177370845e-06, "epoch": 19.050632911392405, "percentage": 95.25, "elapsed_time": "0:10:27", "remaining_time": "0:00:31", "throughput": 2171.23, "total_tokens": 1362832}
 {"current_steps": 3015, "total_steps": 3160, "loss": 0.0001, "lr": 6.488506309255238e-06, "epoch": 19.082278481012658, "percentage": 95.41, "elapsed_time": "0:10:28", "remaining_time": "0:00:30", "throughput": 2171.5, "total_tokens": 1365072}
+{"current_steps": 3020, "total_steps": 3160, "loss": 0.0002, "lr": 6.052581461340411e-06, "epoch": 19.11392405063291, "percentage": 95.57, "elapsed_time": "0:10:29", "remaining_time": "0:00:29", "throughput": 2171.89, "total_tokens": 1367440}
+{"current_steps": 3025, "total_steps": 3160, "loss": 0.0002, "lr": 5.631724772138469e-06, "epoch": 19.145569620253166, "percentage": 95.73, "elapsed_time": "0:10:30", "remaining_time": "0:00:28", "throughput": 2172.21, "total_tokens": 1369744}
+{"current_steps": 3030, "total_steps": 3160, "loss": 0.0009, "lr": 5.225949080131964e-06, "epoch": 19.17721518987342, "percentage": 95.89, "elapsed_time": "0:10:31", "remaining_time": "0:00:27", "throughput": 2172.45, "total_tokens": 1371984}
+{"current_steps": 3035, "total_steps": 3160, "loss": 0.0004, "lr": 4.8352667637490684e-06, "epoch": 19.20886075949367, "percentage": 96.04, "elapsed_time": "0:10:32", "remaining_time": "0:00:26", "throughput": 2172.76, "total_tokens": 1374288}
+{"current_steps": 3040, "total_steps": 3160, "loss": 0.0002, "lr": 4.459689740985206e-06, "epoch": 19.240506329113924, "percentage": 96.2, "elapsed_time": "0:10:33", "remaining_time": "0:00:25", "throughput": 2172.99, "total_tokens": 1376496}
+{"current_steps": 3045, "total_steps": 3160, "loss": 0.0003, "lr": 4.099229469040011e-06, "epoch": 19.272151898734176, "percentage": 96.36, "elapsed_time": "0:10:34", "remaining_time": "0:00:23", "throughput": 2173.25, "total_tokens": 1378736}
+{"current_steps": 3050, "total_steps": 3160, "loss": 0.0003, "lr": 3.7538969439678272e-06, "epoch": 19.303797468354432, "percentage": 96.52, "elapsed_time": "0:10:35", "remaining_time": "0:00:22", "throughput": 2173.47, "total_tokens": 1380944}
+{"current_steps": 3055, "total_steps": 3160, "loss": 0.0001, "lr": 3.423702700341813e-06, "epoch": 19.335443037974684, "percentage": 96.68, "elapsed_time": "0:10:36", "remaining_time": "0:00:21", "throughput": 2173.81, "total_tokens": 1383248}
+{"current_steps": 3060, "total_steps": 3160, "loss": 0.0003, "lr": 3.1086568109331413e-06, "epoch": 19.367088607594937, "percentage": 96.84, "elapsed_time": "0:10:37", "remaining_time": "0:00:20", "throughput": 2174.08, "total_tokens": 1385488}
+{"current_steps": 3065, "total_steps": 3160, "loss": 0.0005, "lr": 2.8087688864033013e-06, "epoch": 19.39873417721519, "percentage": 96.99, "elapsed_time": "0:10:38", "remaining_time": "0:00:19", "throughput": 2174.38, "total_tokens": 1387760}
+{"current_steps": 3070, "total_steps": 3160, "loss": 0.0002, "lr": 2.524048075011165e-06, "epoch": 19.430379746835442, "percentage": 97.15, "elapsed_time": "0:10:39", "remaining_time": "0:00:18", "throughput": 2174.82, "total_tokens": 1390192}
+{"current_steps": 3075, "total_steps": 3160, "loss": 0.0006, "lr": 2.254503062333824e-06, "epoch": 19.462025316455698, "percentage": 97.31, "elapsed_time": "0:10:40", "remaining_time": "0:00:17", "throughput": 2175.1, "total_tokens": 1392464}
+{"current_steps": 3080, "total_steps": 3160, "loss": 0.0001, "lr": 2.0001420710016318e-06, "epoch": 19.49367088607595, "percentage": 97.47, "elapsed_time": "0:10:41", "remaining_time": "0:00:16", "throughput": 2175.44, "total_tokens": 1394768}
+{"current_steps": 3085, "total_steps": 3160, "loss": 0.0005, "lr": 1.7609728604472963e-06, "epoch": 19.525316455696203, "percentage": 97.63, "elapsed_time": "0:10:42", "remaining_time": "0:00:15", "throughput": 2175.79, "total_tokens": 1397072}
+{"current_steps": 3090, "total_steps": 3160, "loss": 0.0003, "lr": 1.5370027266694008e-06, "epoch": 19.556962025316455, "percentage": 97.78, "elapsed_time": "0:10:43", "remaining_time": "0:00:14", "throughput": 2176.06, "total_tokens": 1399344}
+{"current_steps": 3095, "total_steps": 3160, "loss": 0.0002, "lr": 1.3282385020095267e-06, "epoch": 19.588607594936708, "percentage": 97.94, "elapsed_time": "0:10:44", "remaining_time": "0:00:13", "throughput": 2176.27, "total_tokens": 1401552}
+{"current_steps": 3100, "total_steps": 3160, "loss": 0.0003, "lr": 1.1346865549440867e-06, "epoch": 19.620253164556964, "percentage": 98.1, "elapsed_time": "0:10:44", "remaining_time": "0:00:12", "throughput": 2176.46, "total_tokens": 1403760}
+{"current_steps": 3105, "total_steps": 3160, "loss": 0.0001, "lr": 9.563527898899805e-07, "epoch": 19.651898734177216, "percentage": 98.26, "elapsed_time": "0:10:45", "remaining_time": "0:00:11", "throughput": 2176.71, "total_tokens": 1406000}
+{"current_steps": 3110, "total_steps": 3160, "loss": 0.0009, "lr": 7.932426470242948e-07, "epoch": 19.68354430379747, "percentage": 98.42, "elapsed_time": "0:10:46", "remaining_time": "0:00:10", "throughput": 2176.92, "total_tokens": 1408208}
+{"current_steps": 3115, "total_steps": 3160, "loss": 0.0001, "lr": 6.453611021186578e-07, "epoch": 19.71518987341772, "percentage": 98.58, "elapsed_time": "0:10:47", "remaining_time": "0:00:09", "throughput": 2177.13, "total_tokens": 1410416}
+{"current_steps": 3120, "total_steps": 3160, "loss": 0.0001, "lr": 5.12712666387194e-07, "epoch": 19.746835443037973, "percentage": 98.73, "elapsed_time": "0:10:48", "remaining_time": "0:00:08", "throughput": 2177.3, "total_tokens": 1412592}
+{"current_steps": 3125, "total_steps": 3160, "loss": 0.0002, "lr": 3.953013863490784e-07, "epoch": 19.77848101265823, "percentage": 98.89, "elapsed_time": "0:10:49", "remaining_time": "0:00:07", "throughput": 2177.66, "total_tokens": 1414960}
+{"current_steps": 3130, "total_steps": 3160, "loss": 0.0002, "lr": 2.9313084370496955e-07, "epoch": 19.810126582278482, "percentage": 99.05, "elapsed_time": "0:10:50", "remaining_time": "0:00:06", "throughput": 2177.92, "total_tokens": 1417232}
+{"current_steps": 3135, "total_steps": 3160, "loss": 0.0002, "lr": 2.062041552277627e-07, "epoch": 19.841772151898734, "percentage": 99.21, "elapsed_time": "0:10:51", "remaining_time": "0:00:05", "throughput": 2178.28, "total_tokens": 1419568}
+{"current_steps": 3140, "total_steps": 3160, "loss": 0.0003, "lr": 1.3452397266772166e-07, "epoch": 19.873417721518987, "percentage": 99.37, "elapsed_time": "0:10:52", "remaining_time": "0:00:04", "throughput": 2178.45, "total_tokens": 1421776}
+{"current_steps": 3145, "total_steps": 3160, "loss": 0.0008, "lr": 7.809248267121038e-08, "epoch": 19.90506329113924, "percentage": 99.53, "elapsed_time": "0:10:53", "remaining_time": "0:00:03", "throughput": 2178.66, "total_tokens": 1424016}
+{"current_steps": 3150, "total_steps": 3160, "loss": 0.0003, "lr": 3.6911406714246195e-08, "epoch": 19.936708860759495, "percentage": 99.68, "elapsed_time": "0:10:54", "remaining_time": "0:00:02", "throughput": 2178.94, "total_tokens": 1426288}
+{"current_steps": 3155, "total_steps": 3160, "loss": 0.001, "lr": 1.0982001050041657e-08, "epoch": 19.968354430379748, "percentage": 99.84, "elapsed_time": "0:10:55", "remaining_time": "0:00:01", "throughput": 2179.24, "total_tokens": 1428592}
+{"current_steps": 3160, "total_steps": 3160, "loss": 0.0001, "lr": 3.0505667042435294e-10, "epoch": 20.0, "percentage": 100.0, "elapsed_time": "0:10:56", "remaining_time": "0:00:00", "throughput": 2179.18, "total_tokens": 1430592}
+{"current_steps": 3160, "total_steps": 3160, "eval_loss": 0.27992814779281616, "epoch": 20.0, "percentage": 100.0, "elapsed_time": "0:10:58", "remaining_time": "0:00:00", "throughput": 2173.75, "total_tokens": 1430592}
+{"current_steps": 3160, "total_steps": 3160, "epoch": 20.0, "percentage": 100.0, "elapsed_time": "0:10:58", "remaining_time": "0:00:00", "throughput": 2171.03, "total_tokens": 1430592}