Training in progress, step 2260

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:68bebbe95bf7cac99106a7c4bf2dd8be6a8ded99cfc9569c2224480942591b33
+oid sha256:d0bb614352fe25c7ef34dc94ad1b1597177b5794ffd2f7eb1c5a042baa5e54a3
 size 8388736

trainer_log.jsonl

@@ -449,3 +449,25 @@
 {"current_steps": 2150, "total_steps": 2260, "loss": 0.0, "lr": 3.665133433253809e-07, "epoch": 19.02654867256637, "percentage": 95.13, "elapsed_time": "0:03:59", "remaining_time": "0:00:12", "throughput": 2463.99, "total_tokens": 589776}
 {"current_steps": 2155, "total_steps": 2260, "loss": 0.0, "lr": 3.34309935365093e-07, "epoch": 19.07079646017699, "percentage": 95.35, "elapsed_time": "0:03:59", "remaining_time": "0:00:11", "throughput": 2465.23, "total_tokens": 591200}
 {"current_steps": 2160, "total_steps": 2260, "loss": 0.0, "lr": 3.03577582601422e-07, "epoch": 19.115044247787612, "percentage": 95.58, "elapsed_time": "0:04:00", "remaining_time": "0:00:11", "throughput": 2465.78, "total_tokens": 592464}
+{"current_steps": 2165, "total_steps": 2260, "loss": 0.0, "lr": 2.743181179037047e-07, "epoch": 19.15929203539823, "percentage": 95.8, "elapsed_time": "0:04:00", "remaining_time": "0:00:10", "throughput": 2467.0, "total_tokens": 593936}
+{"current_steps": 2170, "total_steps": 2260, "loss": 0.0, "lr": 2.465332862986447e-07, "epoch": 19.20353982300885, "percentage": 96.02, "elapsed_time": "0:04:01", "remaining_time": "0:00:10", "throughput": 2467.78, "total_tokens": 595232}
+{"current_steps": 2175, "total_steps": 2260, "loss": 0.0, "lr": 2.2022474486620427e-07, "epoch": 19.24778761061947, "percentage": 96.24, "elapsed_time": "0:04:01", "remaining_time": "0:00:09", "throughput": 2468.75, "total_tokens": 596592}
+{"current_steps": 2180, "total_steps": 2260, "loss": 0.0, "lr": 1.953940626408024e-07, "epoch": 19.292035398230087, "percentage": 96.46, "elapsed_time": "0:04:02", "remaining_time": "0:00:08", "throughput": 2470.32, "total_tokens": 598176}
+{"current_steps": 2185, "total_steps": 2260, "loss": 0.0001, "lr": 1.720427205177233e-07, "epoch": 19.336283185840706, "percentage": 96.68, "elapsed_time": "0:04:02", "remaining_time": "0:00:08", "throughput": 2472.2, "total_tokens": 599792}
+{"current_steps": 2190, "total_steps": 2260, "loss": 0.0, "lr": 1.5017211116479802e-07, "epoch": 19.38053097345133, "percentage": 96.9, "elapsed_time": "0:04:03", "remaining_time": "0:00:07", "throughput": 2472.97, "total_tokens": 601072}
+{"current_steps": 2195, "total_steps": 2260, "loss": 0.0, "lr": 1.297835389393598e-07, "epoch": 19.424778761061948, "percentage": 97.12, "elapsed_time": "0:04:03", "remaining_time": "0:00:07", "throughput": 2473.61, "total_tokens": 602384}
+{"current_steps": 2200, "total_steps": 2260, "loss": 0.0, "lr": 1.1087821981042856e-07, "epoch": 19.469026548672566, "percentage": 97.35, "elapsed_time": "0:04:03", "remaining_time": "0:00:06", "throughput": 2474.87, "total_tokens": 603840}
+{"current_steps": 2205, "total_steps": 2260, "loss": 0.0, "lr": 9.345728128621611e-08, "epoch": 19.513274336283185, "percentage": 97.57, "elapsed_time": "0:04:04", "remaining_time": "0:00:06", "throughput": 2474.92, "total_tokens": 604944}
+{"current_steps": 2210, "total_steps": 2260, "loss": 0.0, "lr": 7.752176234685771e-08, "epoch": 19.557522123893804, "percentage": 97.79, "elapsed_time": "0:04:04", "remaining_time": "0:00:05", "throughput": 2475.35, "total_tokens": 606224}
+{"current_steps": 2215, "total_steps": 2260, "loss": 0.0, "lr": 6.307261338246718e-08, "epoch": 19.601769911504427, "percentage": 98.01, "elapsed_time": "0:04:05", "remaining_time": "0:00:04", "throughput": 2476.41, "total_tokens": 607632}
+{"current_steps": 2220, "total_steps": 2260, "loss": 0.0, "lr": 5.011069613644892e-08, "epoch": 19.646017699115045, "percentage": 98.23, "elapsed_time": "0:04:05", "remaining_time": "0:00:04", "throughput": 2477.21, "total_tokens": 608960}
+{"current_steps": 2225, "total_steps": 2260, "loss": 0.0, "lr": 3.8636783654100174e-08, "epoch": 19.690265486725664, "percentage": 98.45, "elapsed_time": "0:04:06", "remaining_time": "0:00:03", "throughput": 2478.33, "total_tokens": 610400}
+{"current_steps": 2230, "total_steps": 2260, "loss": 0.0, "lr": 2.865156023650617e-08, "epoch": 19.734513274336283, "percentage": 98.67, "elapsed_time": "0:04:06", "remaining_time": "0:00:03", "throughput": 2478.92, "total_tokens": 611648}
+{"current_steps": 2235, "total_steps": 2260, "loss": 0.0, "lr": 2.0155621399742254e-08, "epoch": 19.778761061946902, "percentage": 98.89, "elapsed_time": "0:04:07", "remaining_time": "0:00:02", "throughput": 2479.46, "total_tokens": 612880}
+{"current_steps": 2240, "total_steps": 2260, "loss": 0.0, "lr": 1.31494738393384e-08, "epoch": 19.82300884955752, "percentage": 99.12, "elapsed_time": "0:04:07", "remaining_time": "0:00:02", "throughput": 2481.1, "total_tokens": 614560}
+{"current_steps": 2245, "total_steps": 2260, "loss": 0.0, "lr": 7.633535400070057e-09, "epoch": 19.86725663716814, "percentage": 99.34, "elapsed_time": "0:04:08", "remaining_time": "0:00:01", "throughput": 2481.61, "total_tokens": 615856}
+{"current_steps": 2250, "total_steps": 2260, "loss": 0.0, "lr": 3.6081350510447365e-09, "epoch": 19.911504424778762, "percentage": 99.56, "elapsed_time": "0:04:08", "remaining_time": "0:00:01", "throughput": 2482.29, "total_tokens": 617136}
+{"current_steps": 2255, "total_steps": 2260, "loss": 0.0, "lr": 1.0735128660649406e-09, "epoch": 19.95575221238938, "percentage": 99.78, "elapsed_time": "0:04:09", "remaining_time": "0:00:00", "throughput": 2483.6, "total_tokens": 618640}
+{"current_steps": 2260, "total_steps": 2260, "loss": 0.0, "lr": 2.982000932294504e-11, "epoch": 20.0, "percentage": 100.0, "elapsed_time": "0:04:09", "remaining_time": "0:00:00", "throughput": 2484.39, "total_tokens": 620240}
+{"current_steps": 2260, "total_steps": 2260, "eval_loss": 0.01861300691962242, "epoch": 20.0, "percentage": 100.0, "elapsed_time": "0:04:10", "remaining_time": "0:00:00", "throughput": 2477.27, "total_tokens": 620240}
+{"current_steps": 2260, "total_steps": 2260, "epoch": 20.0, "percentage": 100.0, "elapsed_time": "0:04:11", "remaining_time": "0:00:00", "throughput": 2464.7, "total_tokens": 620240}