Training in progress, step 39200

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f7ca720b40d7fc27cc18a1f321987a87248aa67a17deef92d4d41644a8b34f71
+oid sha256:dd43b691acd8093caca4884390fe6ea1227fe0595b216975e280e465d819cf1e
 size 460928

trainer_log.jsonl

@@ -7993,3 +7993,44 @@
 {"current_steps": 38995, "total_steps": 40000, "loss": 0.0, "lr": 0.00046796249011277213, "epoch": 276.5632798573975, "percentage": 97.49, "elapsed_time": "4:45:03", "remaining_time": "0:07:20", "throughput": 5820.75, "total_tokens": 99556512}
 {"current_steps": 39000, "total_steps": 40000, "loss": 0.0, "lr": 0.00046332472568669236, "epoch": 276.59893048128345, "percentage": 97.5, "elapsed_time": "4:45:05", "remaining_time": "0:07:18", "throughput": 5820.84, "total_tokens": 99570208}
 {"current_steps": 39000, "total_steps": 40000, "eval_loss": 0.7350195050239563, "epoch": 276.59893048128345, "percentage": 97.5, "elapsed_time": "4:45:09", "remaining_time": "0:07:18", "throughput": 5819.72, "total_tokens": 99570208}
+{"current_steps": 39005, "total_steps": 40000, "loss": 0.0, "lr": 0.0004587100216951578, "epoch": 276.63458110516933, "percentage": 97.51, "elapsed_time": "4:45:12", "remaining_time": "0:07:16", "throughput": 5819.2, "total_tokens": 99582400}
+{"current_steps": 39010, "total_steps": 40000, "loss": 0.0, "lr": 0.00045411837884978265, "epoch": 276.6702317290553, "percentage": 97.52, "elapsed_time": "4:45:14", "remaining_time": "0:07:14", "throughput": 5819.24, "total_tokens": 99594912}
+{"current_steps": 39015, "total_steps": 40000, "loss": 0.0, "lr": 0.00044954979785865045, "epoch": 276.70588235294116, "percentage": 97.54, "elapsed_time": "4:45:16", "remaining_time": "0:07:12", "throughput": 5819.25, "total_tokens": 99606976}
+{"current_steps": 39020, "total_steps": 40000, "loss": 0.0, "lr": 0.00044500427942631426, "epoch": 276.7415329768271, "percentage": 97.55, "elapsed_time": "4:45:18", "remaining_time": "0:07:09", "throughput": 5819.29, "total_tokens": 99619744}
+{"current_steps": 39025, "total_steps": 40000, "loss": 0.0, "lr": 0.0004404818242537467, "epoch": 276.777183600713, "percentage": 97.56, "elapsed_time": "4:45:20", "remaining_time": "0:07:07", "throughput": 5819.32, "total_tokens": 99632288}
+{"current_steps": 39030, "total_steps": 40000, "loss": 0.0, "lr": 0.00043598243303837324, "epoch": 276.81283422459893, "percentage": 97.58, "elapsed_time": "4:45:23", "remaining_time": "0:07:05", "throughput": 5819.38, "total_tokens": 99645280}
+{"current_steps": 39035, "total_steps": 40000, "loss": 0.0, "lr": 0.00043150610647403885, "epoch": 276.8484848484849, "percentage": 97.59, "elapsed_time": "4:45:25", "remaining_time": "0:07:03", "throughput": 5819.4, "total_tokens": 99657440}
+{"current_steps": 39040, "total_steps": 40000, "loss": 0.0, "lr": 0.00042705284525104134, "epoch": 276.88413547237076, "percentage": 97.6, "elapsed_time": "4:45:27", "remaining_time": "0:07:01", "throughput": 5819.48, "total_tokens": 99671232}
+{"current_steps": 39045, "total_steps": 40000, "loss": 0.0, "lr": 0.0004226226500561647, "epoch": 276.9197860962567, "percentage": 97.61, "elapsed_time": "4:45:29", "remaining_time": "0:06:58", "throughput": 5819.58, "total_tokens": 99685152}
+{"current_steps": 39050, "total_steps": 40000, "loss": 0.0, "lr": 0.0004182155215725791, "epoch": 276.9554367201426, "percentage": 97.62, "elapsed_time": "4:45:31", "remaining_time": "0:06:56", "throughput": 5819.56, "total_tokens": 99696448}
+{"current_steps": 39055, "total_steps": 40000, "loss": 0.0, "lr": 0.00041383146047992424, "epoch": 276.99108734402853, "percentage": 97.64, "elapsed_time": "4:45:33", "remaining_time": "0:06:54", "throughput": 5819.61, "total_tokens": 99709280}
+{"current_steps": 39060, "total_steps": 40000, "loss": 0.0, "lr": 0.00040947046745427597, "epoch": 277.02139037433153, "percentage": 97.65, "elapsed_time": "4:45:35", "remaining_time": "0:06:52", "throughput": 5819.49, "total_tokens": 99719464}
+{"current_steps": 39065, "total_steps": 40000, "loss": 0.0, "lr": 0.00040513254316814625, "epoch": 277.0570409982175, "percentage": 97.66, "elapsed_time": "4:45:37", "remaining_time": "0:06:50", "throughput": 5819.51, "total_tokens": 99731496}
+{"current_steps": 39070, "total_steps": 40000, "loss": 0.0, "lr": 0.0004008176882905168, "epoch": 277.09269162210336, "percentage": 97.67, "elapsed_time": "4:45:39", "remaining_time": "0:06:47", "throughput": 5819.5, "total_tokens": 99743112}
+{"current_steps": 39075, "total_steps": 40000, "loss": 0.0, "lr": 0.00039652590348677184, "epoch": 277.1283422459893, "percentage": 97.69, "elapsed_time": "4:45:41", "remaining_time": "0:06:45", "throughput": 5819.57, "total_tokens": 99756648}
+{"current_steps": 39080, "total_steps": 40000, "loss": 0.0, "lr": 0.00039225718941878206, "epoch": 277.16399286987524, "percentage": 97.7, "elapsed_time": "4:45:43", "remaining_time": "0:06:43", "throughput": 5819.65, "total_tokens": 99770088}
+{"current_steps": 39085, "total_steps": 40000, "loss": 0.0, "lr": 0.00038801154674480417, "epoch": 277.19964349376113, "percentage": 97.71, "elapsed_time": "4:45:45", "remaining_time": "0:06:41", "throughput": 5819.67, "total_tokens": 99782664}
+{"current_steps": 39090, "total_steps": 40000, "loss": 0.0, "lr": 0.00038378897611959784, "epoch": 277.2352941176471, "percentage": 97.72, "elapsed_time": "4:45:47", "remaining_time": "0:06:39", "throughput": 5819.76, "total_tokens": 99796488}
+{"current_steps": 39095, "total_steps": 40000, "loss": 0.0, "lr": 0.00037958947819430875, "epoch": 277.27094474153296, "percentage": 97.74, "elapsed_time": "4:45:49", "remaining_time": "0:06:37", "throughput": 5819.86, "total_tokens": 99810504}
+{"current_steps": 39100, "total_steps": 40000, "loss": 0.0, "lr": 0.0003754130536165856, "epoch": 277.3065953654189, "percentage": 97.75, "elapsed_time": "4:45:52", "remaining_time": "0:06:34", "throughput": 5819.98, "total_tokens": 99824744}
+{"current_steps": 39105, "total_steps": 40000, "loss": 0.0, "lr": 0.0003712597030304632, "epoch": 277.3422459893048, "percentage": 97.76, "elapsed_time": "4:45:54", "remaining_time": "0:06:32", "throughput": 5820.03, "total_tokens": 99837704}
+{"current_steps": 39110, "total_steps": 40000, "loss": 0.0, "lr": 0.00036712942707646247, "epoch": 277.3778966131907, "percentage": 97.78, "elapsed_time": "4:45:56", "remaining_time": "0:06:30", "throughput": 5820.1, "total_tokens": 99850888}
+{"current_steps": 39115, "total_steps": 40000, "loss": 0.0, "lr": 0.00036302222639149063, "epoch": 277.41354723707667, "percentage": 97.79, "elapsed_time": "4:45:58", "remaining_time": "0:06:28", "throughput": 5820.11, "total_tokens": 99862824}
+{"current_steps": 39120, "total_steps": 40000, "loss": 0.0, "lr": 0.000358938101608941, "epoch": 277.44919786096256, "percentage": 97.8, "elapsed_time": "4:46:00", "remaining_time": "0:06:26", "throughput": 5820.19, "total_tokens": 99876520}
+{"current_steps": 39125, "total_steps": 40000, "loss": 0.0, "lr": 0.0003548770533586598, "epoch": 277.4848484848485, "percentage": 97.81, "elapsed_time": "4:46:02", "remaining_time": "0:06:23", "throughput": 5820.25, "total_tokens": 99889736}
+{"current_steps": 39130, "total_steps": 40000, "loss": 0.0, "lr": 0.0003508390822668961, "epoch": 277.5204991087344, "percentage": 97.82, "elapsed_time": "4:46:04", "remaining_time": "0:06:21", "throughput": 5820.27, "total_tokens": 99902088}
+{"current_steps": 39135, "total_steps": 40000, "loss": 0.0, "lr": 0.00034682418895633503, "epoch": 277.5561497326203, "percentage": 97.84, "elapsed_time": "4:46:06", "remaining_time": "0:06:19", "throughput": 5820.32, "total_tokens": 99914760}
+{"current_steps": 39140, "total_steps": 40000, "loss": 0.0, "lr": 0.0003428323740461647, "epoch": 277.5918003565062, "percentage": 97.85, "elapsed_time": "4:46:08", "remaining_time": "0:06:17", "throughput": 5820.28, "total_tokens": 99925704}
+{"current_steps": 39145, "total_steps": 40000, "loss": 0.0, "lr": 0.00033886363815194276, "epoch": 277.62745098039215, "percentage": 97.86, "elapsed_time": "4:46:10", "remaining_time": "0:06:15", "throughput": 5820.33, "total_tokens": 99938504}
+{"current_steps": 39150, "total_steps": 40000, "loss": 0.0, "lr": 0.0003349179818857129, "epoch": 277.6631016042781, "percentage": 97.88, "elapsed_time": "4:46:12", "remaining_time": "0:06:12", "throughput": 5820.44, "total_tokens": 99952840}
+{"current_steps": 39155, "total_steps": 40000, "loss": 0.0, "lr": 0.0003309954058559383, "epoch": 277.698752228164, "percentage": 97.89, "elapsed_time": "4:46:14", "remaining_time": "0:06:10", "throughput": 5820.42, "total_tokens": 99964040}
+{"current_steps": 39160, "total_steps": 40000, "loss": 0.0, "lr": 0.0003270959106675186, "epoch": 277.7344028520499, "percentage": 97.9, "elapsed_time": "4:46:16", "remaining_time": "0:06:08", "throughput": 5820.46, "total_tokens": 99976840}
+{"current_steps": 39165, "total_steps": 40000, "loss": 0.0, "lr": 0.0003232194969218227, "epoch": 277.7700534759358, "percentage": 97.91, "elapsed_time": "4:46:18", "remaining_time": "0:06:06", "throughput": 5820.49, "total_tokens": 99989416}
+{"current_steps": 39170, "total_steps": 40000, "loss": 0.0, "lr": 0.00031936616521663905, "epoch": 277.80570409982175, "percentage": 97.92, "elapsed_time": "4:46:20", "remaining_time": "0:06:04", "throughput": 5820.53, "total_tokens": 100002120}
+{"current_steps": 39175, "total_steps": 40000, "loss": 0.0, "lr": 0.00031553591614619236, "epoch": 277.84135472370764, "percentage": 97.94, "elapsed_time": "4:46:22", "remaining_time": "0:06:01", "throughput": 5820.53, "total_tokens": 100013864}
+{"current_steps": 39180, "total_steps": 40000, "loss": 0.0, "lr": 0.00031172875030117676, "epoch": 277.8770053475936, "percentage": 97.95, "elapsed_time": "4:46:25", "remaining_time": "0:05:59", "throughput": 5820.57, "total_tokens": 100026600}
+{"current_steps": 39185, "total_steps": 40000, "loss": 0.0, "lr": 0.0003079446682686726, "epoch": 277.9126559714795, "percentage": 97.96, "elapsed_time": "4:46:27", "remaining_time": "0:05:57", "throughput": 5820.62, "total_tokens": 100039368}
+{"current_steps": 39190, "total_steps": 40000, "loss": 0.0, "lr": 0.0003041836706322465, "epoch": 277.9483065953654, "percentage": 97.97, "elapsed_time": "4:46:29", "remaining_time": "0:05:55", "throughput": 5820.72, "total_tokens": 100053384}
+{"current_steps": 39195, "total_steps": 40000, "loss": 0.0, "lr": 0.0003004457579719011, "epoch": 277.98395721925135, "percentage": 97.99, "elapsed_time": "4:46:31", "remaining_time": "0:05:53", "throughput": 5820.81, "total_tokens": 100067176}
+{"current_steps": 39200, "total_steps": 40000, "loss": 0.0, "lr": 0.00029673093086405867, "epoch": 278.01426024955435, "percentage": 98.0, "elapsed_time": "4:46:33", "remaining_time": "0:05:50", "throughput": 5820.69, "total_tokens": 100077240}
+{"current_steps": 39200, "total_steps": 40000, "eval_loss": 0.7423530220985413, "epoch": 278.01426024955435, "percentage": 98.0, "elapsed_time": "4:46:36", "remaining_time": "0:05:50", "throughput": 5819.58, "total_tokens": 100077240}