Training in progress, step 39800

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0a2af390a171906445dfd051c096228d6406eda6c13aab63e5e5413af697b3a2
+oid sha256:7b050cdd53e0a7410059025da061cf9fbff1af6af7967a4501ba605b0212589c
 size 460928

trainer_log.jsonl

@@ -8116,3 +8116,44 @@
 {"current_steps": 39595, "total_steps": 40000, "loss": 0.0007, "lr": 7.625288752117209e-05, "epoch": 122.20710973724884, "percentage": 98.99, "elapsed_time": "4:49:44", "remaining_time": "0:02:57", "throughput": 3478.3, "total_tokens": 60468912}
 {"current_steps": 39600, "total_steps": 40000, "loss": 0.0008, "lr": 7.4386456836667e-05, "epoch": 122.2225656877898, "percentage": 99.0, "elapsed_time": "4:49:46", "remaining_time": "0:02:55", "throughput": 3478.35, "total_tokens": 60476592}
 {"current_steps": 39600, "total_steps": 40000, "eval_loss": 1.4520829916000366, "epoch": 122.2225656877898, "percentage": 99.0, "elapsed_time": "4:49:53", "remaining_time": "0:02:55", "throughput": 3476.94, "total_tokens": 60476592}
+{"current_steps": 39605, "total_steps": 40000, "loss": 0.0008, "lr": 7.254314656586214e-05, "epoch": 122.23802163833076, "percentage": 99.01, "elapsed_time": "4:49:57", "remaining_time": "0:02:53", "throughput": 3476.67, "total_tokens": 60484400}
+{"current_steps": 39610, "total_steps": 40000, "loss": 0.001, "lr": 7.07229569929968e-05, "epoch": 122.25347758887172, "percentage": 99.02, "elapsed_time": "4:49:59", "remaining_time": "0:02:51", "throughput": 3476.72, "total_tokens": 60492176}
+{"current_steps": 39615, "total_steps": 40000, "loss": 0.0008, "lr": 6.892588839879643e-05, "epoch": 122.26893353941267, "percentage": 99.04, "elapsed_time": "4:50:01", "remaining_time": "0:02:49", "throughput": 3476.79, "total_tokens": 60500272}
+{"current_steps": 39620, "total_steps": 40000, "loss": 0.0007, "lr": 6.71519410603727e-05, "epoch": 122.28438948995363, "percentage": 99.05, "elapsed_time": "4:50:03", "remaining_time": "0:02:46", "throughput": 3476.78, "total_tokens": 60507888}
+{"current_steps": 39625, "total_steps": 40000, "loss": 0.0009, "lr": 6.540111525129011e-05, "epoch": 122.2998454404946, "percentage": 99.06, "elapsed_time": "4:50:05", "remaining_time": "0:02:44", "throughput": 3476.85, "total_tokens": 60515920}
+{"current_steps": 39630, "total_steps": 40000, "loss": 0.001, "lr": 6.367341124154934e-05, "epoch": 122.31530139103555, "percentage": 99.08, "elapsed_time": "4:50:07", "remaining_time": "0:02:42", "throughput": 3476.86, "total_tokens": 60523056}
+{"current_steps": 39635, "total_steps": 40000, "loss": 0.0006, "lr": 6.19688292975873e-05, "epoch": 122.3307573415765, "percentage": 99.09, "elapsed_time": "4:50:09", "remaining_time": "0:02:40", "throughput": 3476.91, "total_tokens": 60530672}
+{"current_steps": 39640, "total_steps": 40000, "loss": 0.0008, "lr": 6.0287369682260336e-05, "epoch": 122.34621329211747, "percentage": 99.1, "elapsed_time": "4:50:11", "remaining_time": "0:02:38", "throughput": 3476.93, "total_tokens": 60537936}
+{"current_steps": 39645, "total_steps": 40000, "loss": 0.0007, "lr": 5.8629032654894384e-05, "epoch": 122.36166924265842, "percentage": 99.11, "elapsed_time": "4:50:13", "remaining_time": "0:02:35", "throughput": 3476.97, "total_tokens": 60545552}
+{"current_steps": 39650, "total_steps": 40000, "loss": 0.0009, "lr": 5.699381847120155e-05, "epoch": 122.37712519319938, "percentage": 99.12, "elapsed_time": "4:50:15", "remaining_time": "0:02:33", "throughput": 3477.04, "total_tokens": 60553552}
+{"current_steps": 39655, "total_steps": 40000, "loss": 0.0008, "lr": 5.5381727383380094e-05, "epoch": 122.39258114374034, "percentage": 99.14, "elapsed_time": "4:50:17", "remaining_time": "0:02:31", "throughput": 3477.07, "total_tokens": 60560944}
+{"current_steps": 39660, "total_steps": 40000, "loss": 0.0013, "lr": 5.379275964001451e-05, "epoch": 122.4080370942813, "percentage": 99.15, "elapsed_time": "4:50:19", "remaining_time": "0:02:29", "throughput": 3477.11, "total_tokens": 60568432}
+{"current_steps": 39665, "total_steps": 40000, "loss": 0.0009, "lr": 5.222691548614211e-05, "epoch": 122.42349304482225, "percentage": 99.16, "elapsed_time": "4:50:21", "remaining_time": "0:02:27", "throughput": 3477.18, "total_tokens": 60576592}
+{"current_steps": 39670, "total_steps": 40000, "loss": 0.0008, "lr": 5.068419516323641e-05, "epoch": 122.43894899536322, "percentage": 99.17, "elapsed_time": "4:50:23", "remaining_time": "0:02:24", "throughput": 3477.22, "total_tokens": 60584208}
+{"current_steps": 39675, "total_steps": 40000, "loss": 0.0007, "lr": 4.91645989092071e-05, "epoch": 122.45440494590417, "percentage": 99.19, "elapsed_time": "4:50:25", "remaining_time": "0:02:22", "throughput": 3477.29, "total_tokens": 60592144}
+{"current_steps": 39680, "total_steps": 40000, "loss": 0.0006, "lr": 4.7668126958400056e-05, "epoch": 122.46986089644513, "percentage": 99.2, "elapsed_time": "4:50:27", "remaining_time": "0:02:20", "throughput": 3477.34, "total_tokens": 60600048}
+{"current_steps": 39685, "total_steps": 40000, "loss": 0.001, "lr": 4.619477954159734e-05, "epoch": 122.4853168469861, "percentage": 99.21, "elapsed_time": "4:50:29", "remaining_time": "0:02:18", "throughput": 3477.41, "total_tokens": 60608016}
+{"current_steps": 39690, "total_steps": 40000, "loss": 0.0005, "lr": 4.4744556885983884e-05, "epoch": 122.50077279752705, "percentage": 99.22, "elapsed_time": "4:50:31", "remaining_time": "0:02:16", "throughput": 3477.46, "total_tokens": 60615728}
+{"current_steps": 39695, "total_steps": 40000, "loss": 0.0007, "lr": 4.331745921523078e-05, "epoch": 122.516228748068, "percentage": 99.24, "elapsed_time": "4:50:33", "remaining_time": "0:02:13", "throughput": 3477.49, "total_tokens": 60623120}
+{"current_steps": 39700, "total_steps": 40000, "loss": 0.0005, "lr": 4.191348674937867e-05, "epoch": 122.53168469860897, "percentage": 99.25, "elapsed_time": "4:50:34", "remaining_time": "0:02:11", "throughput": 3477.53, "total_tokens": 60630800}
+{"current_steps": 39705, "total_steps": 40000, "loss": 0.0008, "lr": 4.0532639704971006e-05, "epoch": 122.54714064914992, "percentage": 99.26, "elapsed_time": "4:50:36", "remaining_time": "0:02:09", "throughput": 3477.6, "total_tokens": 60638800}
+{"current_steps": 39710, "total_steps": 40000, "loss": 0.0011, "lr": 3.917491829493747e-05, "epoch": 122.56259659969088, "percentage": 99.28, "elapsed_time": "4:50:38", "remaining_time": "0:02:07", "throughput": 3477.64, "total_tokens": 60646352}
+{"current_steps": 39715, "total_steps": 40000, "loss": 0.0017, "lr": 3.78403227286439e-05, "epoch": 122.57805255023185, "percentage": 99.29, "elapsed_time": "4:50:40", "remaining_time": "0:02:05", "throughput": 3477.67, "total_tokens": 60653712}
+{"current_steps": 39720, "total_steps": 40000, "loss": 0.0007, "lr": 3.652885321192567e-05, "epoch": 122.5935085007728, "percentage": 99.3, "elapsed_time": "4:50:42", "remaining_time": "0:02:02", "throughput": 3477.72, "total_tokens": 60661520}
+{"current_steps": 39725, "total_steps": 40000, "loss": 0.0006, "lr": 3.524050994702099e-05, "epoch": 122.60896445131375, "percentage": 99.31, "elapsed_time": "4:50:44", "remaining_time": "0:02:00", "throughput": 3477.8, "total_tokens": 60669680}
+{"current_steps": 39730, "total_steps": 40000, "loss": 0.0007, "lr": 3.3975293132604276e-05, "epoch": 122.62442040185472, "percentage": 99.33, "elapsed_time": "4:50:46", "remaining_time": "0:01:58", "throughput": 3477.87, "total_tokens": 60677744}
+{"current_steps": 39735, "total_steps": 40000, "loss": 0.0007, "lr": 3.2733202963786125e-05, "epoch": 122.63987635239567, "percentage": 99.34, "elapsed_time": "4:50:48", "remaining_time": "0:01:56", "throughput": 3477.9, "total_tokens": 60685264}
+{"current_steps": 39740, "total_steps": 40000, "loss": 0.0006, "lr": 3.15142396321133e-05, "epoch": 122.65533230293663, "percentage": 99.35, "elapsed_time": "4:50:50", "remaining_time": "0:01:54", "throughput": 3477.93, "total_tokens": 60692528}
+{"current_steps": 39745, "total_steps": 40000, "loss": 0.0013, "lr": 3.0318403325552132e-05, "epoch": 122.67078825347758, "percentage": 99.36, "elapsed_time": "4:50:52", "remaining_time": "0:01:51", "throughput": 3477.97, "total_tokens": 60700048}
+{"current_steps": 39750, "total_steps": 40000, "loss": 0.0008, "lr": 2.914569422855506e-05, "epoch": 122.68624420401855, "percentage": 99.38, "elapsed_time": "4:50:54", "remaining_time": "0:01:49", "throughput": 3478.02, "total_tokens": 60707888}
+{"current_steps": 39755, "total_steps": 40000, "loss": 0.0009, "lr": 2.7996112521927462e-05, "epoch": 122.7017001545595, "percentage": 99.39, "elapsed_time": "4:50:56", "remaining_time": "0:01:47", "throughput": 3478.04, "total_tokens": 60715120}
+{"current_steps": 39760, "total_steps": 40000, "loss": 0.0005, "lr": 2.68696583829775e-05, "epoch": 122.71715610510046, "percentage": 99.4, "elapsed_time": "4:50:58", "remaining_time": "0:01:45", "throughput": 3478.07, "total_tokens": 60722416}
+{"current_steps": 39765, "total_steps": 40000, "loss": 0.0007, "lr": 2.576633198539957e-05, "epoch": 122.73261205564143, "percentage": 99.41, "elapsed_time": "4:51:00", "remaining_time": "0:01:43", "throughput": 3478.14, "total_tokens": 60730512}
+{"current_steps": 39770, "total_steps": 40000, "loss": 0.0005, "lr": 2.46861334993409e-05, "epoch": 122.74806800618238, "percentage": 99.42, "elapsed_time": "4:51:02", "remaining_time": "0:01:40", "throughput": 3478.15, "total_tokens": 60737488}
+{"current_steps": 39775, "total_steps": 40000, "loss": 0.0009, "lr": 2.3629063091384903e-05, "epoch": 122.76352395672333, "percentage": 99.44, "elapsed_time": "4:51:04", "remaining_time": "0:01:38", "throughput": 3478.17, "total_tokens": 60744784}
+{"current_steps": 39780, "total_steps": 40000, "loss": 0.001, "lr": 2.2595120924567834e-05, "epoch": 122.7789799072643, "percentage": 99.45, "elapsed_time": "4:51:06", "remaining_time": "0:01:36", "throughput": 3478.2, "total_tokens": 60752112}
+{"current_steps": 39785, "total_steps": 40000, "loss": 0.0006, "lr": 2.158430715829551e-05, "epoch": 122.79443585780525, "percentage": 99.46, "elapsed_time": "4:51:08", "remaining_time": "0:01:34", "throughput": 3478.24, "total_tokens": 60759664}
+{"current_steps": 39790, "total_steps": 40000, "loss": 0.0006, "lr": 2.059662194849321e-05, "epoch": 122.80989180834621, "percentage": 99.48, "elapsed_time": "4:51:10", "remaining_time": "0:01:32", "throughput": 3478.29, "total_tokens": 60767344}
+{"current_steps": 39795, "total_steps": 40000, "loss": 0.0006, "lr": 1.9632065447422463e-05, "epoch": 122.82534775888718, "percentage": 99.49, "elapsed_time": "4:51:12", "remaining_time": "0:01:30", "throughput": 3478.33, "total_tokens": 60774960}
+{"current_steps": 39800, "total_steps": 40000, "loss": 0.0005, "lr": 1.8690637803880916e-05, "epoch": 122.84080370942813, "percentage": 99.5, "elapsed_time": "4:51:14", "remaining_time": "0:01:27", "throughput": 3478.4, "total_tokens": 60782960}
+{"current_steps": 39800, "total_steps": 40000, "eval_loss": 1.4524598121643066, "epoch": 122.84080370942813, "percentage": 99.5, "elapsed_time": "4:51:21", "remaining_time": "0:01:27", "throughput": 3476.99, "total_tokens": 60782960}