Training in progress, step 1500

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a225bd913724c70fcf37224d6ed12ca0b7f496af35bb748eaa2f9a007d3ef7de
+oid sha256:dd8ef12cc5b3a9dac373a83ff3eb524d9fc49e7cc71124d1017c3864c7ea4752
 size 83945296

trainer_log.jsonl

@@ -100,3 +100,54 @@
 {"current_steps": 990, "total_steps": 1854, "loss": 0.8135, "accuracy": 0.59375, "learning_rate": 2.2338200545580577e-06, "epoch": 1.600323297635886, "percentage": 53.4, "elapsed_time": "2:39:56", "remaining_time": "2:19:34"}
 {"current_steps": 1000, "total_steps": 1854, "loss": 0.7571, "accuracy": 0.606249988079071, "learning_rate": 2.191736455761947e-06, "epoch": 1.616488179430188, "percentage": 53.94, "elapsed_time": "2:41:26", "remaining_time": "2:17:52"}
 {"current_steps": 1000, "total_steps": 1854, "eval_loss": 0.8679323792457581, "epoch": 1.616488179430188, "percentage": 53.94, "elapsed_time": "2:44:41", "remaining_time": "2:20:38"}
+{"current_steps": 1010, "total_steps": 1854, "loss": 0.8393, "accuracy": 0.625, "learning_rate": 2.1497413764574673e-06, "epoch": 1.6326530612244898, "percentage": 54.48, "elapsed_time": "2:46:27", "remaining_time": "2:19:06"}
+{"current_steps": 1020, "total_steps": 1854, "loss": 0.8035, "accuracy": 0.5874999761581421, "learning_rate": 2.1078468757516395e-06, "epoch": 1.6488179430187917, "percentage": 55.02, "elapsed_time": "2:48:00", "remaining_time": "2:17:22"}
+{"current_steps": 1030, "total_steps": 1854, "loss": 0.832, "accuracy": 0.574999988079071, "learning_rate": 2.0660649838698145e-06, "epoch": 1.6649828248130936, "percentage": 55.56, "elapsed_time": "2:49:35", "remaining_time": "2:15:40"}
+{"current_steps": 1040, "total_steps": 1854, "loss": 0.8888, "accuracy": 0.606249988079071, "learning_rate": 2.0244076987011284e-06, "epoch": 1.6811477066073954, "percentage": 56.09, "elapsed_time": "2:51:11", "remaining_time": "2:13:59"}
+{"current_steps": 1050, "total_steps": 1854, "loss": 0.8526, "accuracy": 0.5562499761581421, "learning_rate": 1.982886982353251e-06, "epoch": 1.6973125884016973, "percentage": 56.63, "elapsed_time": "2:52:46", "remaining_time": "2:12:17"}
+{"current_steps": 1060, "total_steps": 1854, "loss": 0.9022, "accuracy": 0.643750011920929, "learning_rate": 1.941514757717392e-06, "epoch": 1.7134774701959992, "percentage": 57.17, "elapsed_time": "2:54:24", "remaining_time": "2:10:38"}
+{"current_steps": 1070, "total_steps": 1854, "loss": 0.8692, "accuracy": 0.5562499761581421, "learning_rate": 1.9003029050445953e-06, "epoch": 1.729642351990301, "percentage": 57.71, "elapsed_time": "2:56:04", "remaining_time": "2:09:00"}
+{"current_steps": 1080, "total_steps": 1854, "loss": 0.8306, "accuracy": 0.5874999761581421, "learning_rate": 1.8592632585342523e-06, "epoch": 1.745807233784603, "percentage": 58.25, "elapsed_time": "2:57:38", "remaining_time": "2:07:18"}
+{"current_steps": 1090, "total_steps": 1854, "loss": 0.8272, "accuracy": 0.581250011920929, "learning_rate": 1.8184076029358527e-06, "epoch": 1.7619721155789048, "percentage": 58.79, "elapsed_time": "2:59:11", "remaining_time": "2:05:35"}
+{"current_steps": 1100, "total_steps": 1854, "loss": 0.8388, "accuracy": 0.5687500238418579, "learning_rate": 1.7777476701649318e-06, "epoch": 1.7781369973732066, "percentage": 59.33, "elapsed_time": "3:00:49", "remaining_time": "2:03:56"}
+{"current_steps": 1110, "total_steps": 1854, "loss": 0.814, "accuracy": 0.5375000238418579, "learning_rate": 1.7372951359341925e-06, "epoch": 1.7943018791675085, "percentage": 59.87, "elapsed_time": "3:02:20", "remaining_time": "2:02:13"}
+{"current_steps": 1120, "total_steps": 1854, "loss": 0.801, "accuracy": 0.625, "learning_rate": 1.6970616164007547e-06, "epoch": 1.8104667609618104, "percentage": 60.41, "elapsed_time": "3:03:47", "remaining_time": "2:00:27"}
+{"current_steps": 1130, "total_steps": 1854, "loss": 0.8591, "accuracy": 0.5625, "learning_rate": 1.6570586648305276e-06, "epoch": 1.8266316427561122, "percentage": 60.95, "elapsed_time": "3:05:21", "remaining_time": "1:58:45"}
+{"current_steps": 1140, "total_steps": 1854, "loss": 0.8453, "accuracy": 0.6000000238418579, "learning_rate": 1.6172977682806151e-06, "epoch": 1.842796524550414, "percentage": 61.49, "elapsed_time": "3:06:53", "remaining_time": "1:57:03"}
+{"current_steps": 1150, "total_steps": 1854, "loss": 0.8601, "accuracy": 0.581250011920929, "learning_rate": 1.5777903443007586e-06, "epoch": 1.858961406344716, "percentage": 62.03, "elapsed_time": "3:08:27", "remaining_time": "1:55:22"}
+{"current_steps": 1160, "total_steps": 1854, "loss": 0.8417, "accuracy": 0.643750011920929, "learning_rate": 1.5385477376547226e-06, "epoch": 1.8751262881390178, "percentage": 62.57, "elapsed_time": "3:10:03", "remaining_time": "1:53:42"}
+{"current_steps": 1170, "total_steps": 1854, "loss": 0.8383, "accuracy": 0.612500011920929, "learning_rate": 1.4995812170625845e-06, "epoch": 1.89129116993332, "percentage": 63.11, "elapsed_time": "3:11:39", "remaining_time": "1:52:02"}
+{"current_steps": 1180, "total_steps": 1854, "loss": 0.8522, "accuracy": 0.625, "learning_rate": 1.4609019719648666e-06, "epoch": 1.9074560517276218, "percentage": 63.65, "elapsed_time": "3:13:16", "remaining_time": "1:50:23"}
+{"current_steps": 1190, "total_steps": 1854, "loss": 0.7723, "accuracy": 0.612500011920929, "learning_rate": 1.42252110930943e-06, "epoch": 1.9236209335219236, "percentage": 64.19, "elapsed_time": "3:14:45", "remaining_time": "1:48:40"}
+{"current_steps": 1200, "total_steps": 1854, "loss": 0.8606, "accuracy": 0.5687500238418579, "learning_rate": 1.3844496503620493e-06, "epoch": 1.9397858153162255, "percentage": 64.72, "elapsed_time": "3:16:24", "remaining_time": "1:47:02"}
+{"current_steps": 1210, "total_steps": 1854, "loss": 0.9234, "accuracy": 0.581250011920929, "learning_rate": 1.3466985275416081e-06, "epoch": 1.9559506971105274, "percentage": 65.26, "elapsed_time": "3:18:02", "remaining_time": "1:45:24"}
+{"current_steps": 1220, "total_steps": 1854, "loss": 0.7986, "accuracy": 0.65625, "learning_rate": 1.309278581280791e-06, "epoch": 1.9721155789048292, "percentage": 65.8, "elapsed_time": "3:19:35", "remaining_time": "1:43:43"}
+{"current_steps": 1230, "total_steps": 1854, "loss": 0.8795, "accuracy": 0.581250011920929, "learning_rate": 1.272200556913199e-06, "epoch": 1.9882804606991311, "percentage": 66.34, "elapsed_time": "3:21:11", "remaining_time": "1:42:04"}
+{"current_steps": 1240, "total_steps": 1854, "loss": 0.86, "accuracy": 0.5874999761581421, "learning_rate": 1.2354751015877698e-06, "epoch": 2.004445342493433, "percentage": 66.88, "elapsed_time": "3:22:43", "remaining_time": "1:40:22"}
+{"current_steps": 1250, "total_steps": 1854, "loss": 0.8366, "accuracy": 0.625, "learning_rate": 1.1991127612113945e-06, "epoch": 2.020610224287735, "percentage": 67.42, "elapsed_time": "3:24:19", "remaining_time": "1:38:43"}
+{"current_steps": 1260, "total_steps": 1854, "loss": 0.8249, "accuracy": 0.5874999761581421, "learning_rate": 1.1631239774206035e-06, "epoch": 2.036775106082037, "percentage": 67.96, "elapsed_time": "3:25:50", "remaining_time": "1:37:02"}
+{"current_steps": 1270, "total_steps": 1854, "loss": 0.7858, "accuracy": 0.625, "learning_rate": 1.1275190845831978e-06, "epoch": 2.052939987876339, "percentage": 68.5, "elapsed_time": "3:27:29", "remaining_time": "1:35:25"}
+{"current_steps": 1280, "total_steps": 1854, "loss": 0.8236, "accuracy": 0.612500011920929, "learning_rate": 1.0923083068306778e-06, "epoch": 2.0691048696706407, "percentage": 69.04, "elapsed_time": "3:29:08", "remaining_time": "1:33:47"}
+{"current_steps": 1290, "total_steps": 1854, "loss": 0.7958, "accuracy": 0.606249988079071, "learning_rate": 1.0575017551223348e-06, "epoch": 2.0852697514649425, "percentage": 69.58, "elapsed_time": "3:30:37", "remaining_time": "1:32:05"}
+{"current_steps": 1300, "total_steps": 1854, "loss": 0.8376, "accuracy": 0.606249988079071, "learning_rate": 1.023109424341833e-06, "epoch": 2.1014346332592444, "percentage": 70.12, "elapsed_time": "3:32:13", "remaining_time": "1:30:26"}
+{"current_steps": 1310, "total_steps": 1854, "loss": 0.8093, "accuracy": 0.6312500238418579, "learning_rate": 9.891411904271273e-07, "epoch": 2.1175995150535463, "percentage": 70.66, "elapsed_time": "3:33:48", "remaining_time": "1:28:47"}
+{"current_steps": 1320, "total_steps": 1854, "loss": 0.8271, "accuracy": 0.59375, "learning_rate": 9.556068075345363e-07, "epoch": 2.133764396847848, "percentage": 71.2, "elapsed_time": "3:35:23", "remaining_time": "1:27:08"}
+{"current_steps": 1330, "total_steps": 1854, "loss": 0.8554, "accuracy": 0.581250011920929, "learning_rate": 9.225159052377838e-07, "epoch": 2.14992927864215, "percentage": 71.74, "elapsed_time": "3:36:59", "remaining_time": "1:25:29"}
+{"current_steps": 1340, "total_steps": 1854, "loss": 0.7449, "accuracy": 0.581250011920929, "learning_rate": 8.898779857628184e-07, "epoch": 2.166094160436452, "percentage": 72.28, "elapsed_time": "3:38:33", "remaining_time": "1:23:50"}
+{"current_steps": 1350, "total_steps": 1854, "loss": 0.862, "accuracy": 0.5625, "learning_rate": 8.577024212591975e-07, "epoch": 2.1822590422307537, "percentage": 72.82, "elapsed_time": "3:40:08", "remaining_time": "1:22:11"}
+{"current_steps": 1360, "total_steps": 1854, "loss": 0.8505, "accuracy": 0.581250011920929, "learning_rate": 8.259984511088276e-07, "epoch": 2.1984239240250556, "percentage": 73.35, "elapsed_time": "3:41:42", "remaining_time": "1:20:32"}
+{"current_steps": 1370, "total_steps": 1854, "loss": 0.8275, "accuracy": 0.637499988079071, "learning_rate": 7.947751792728237e-07, "epoch": 2.2145888058193575, "percentage": 73.89, "elapsed_time": "3:43:14", "remaining_time": "1:18:52"}
+{"current_steps": 1380, "total_steps": 1854, "loss": 0.8546, "accuracy": 0.606249988079071, "learning_rate": 7.640415716772626e-07, "epoch": 2.2307536876136593, "percentage": 74.43, "elapsed_time": "3:44:55", "remaining_time": "1:17:15"}
+{"current_steps": 1390, "total_steps": 1854, "loss": 0.8078, "accuracy": 0.643750011920929, "learning_rate": 7.338064536385722e-07, "epoch": 2.246918569407961, "percentage": 74.97, "elapsed_time": "3:46:33", "remaining_time": "1:15:37"}
+{"current_steps": 1400, "total_steps": 1854, "loss": 0.9119, "accuracy": 0.581250011920929, "learning_rate": 7.040785073292883e-07, "epoch": 2.263083451202263, "percentage": 75.51, "elapsed_time": "3:48:07", "remaining_time": "1:13:58"}
+{"current_steps": 1410, "total_steps": 1854, "loss": 0.7679, "accuracy": 0.643750011920929, "learning_rate": 6.748662692849297e-07, "epoch": 2.279248332996565, "percentage": 76.05, "elapsed_time": "3:49:39", "remaining_time": "1:12:19"}
+{"current_steps": 1420, "total_steps": 1854, "loss": 0.8073, "accuracy": 0.6000000238418579, "learning_rate": 6.46178127952686e-07, "epoch": 2.295413214790867, "percentage": 76.59, "elapsed_time": "3:51:13", "remaining_time": "1:10:40"}
+{"current_steps": 1430, "total_steps": 1854, "loss": 0.8196, "accuracy": 0.59375, "learning_rate": 6.180223212826289e-07, "epoch": 2.3115780965851687, "percentage": 77.13, "elapsed_time": "3:52:46", "remaining_time": "1:09:01"}
+{"current_steps": 1440, "total_steps": 1854, "loss": 0.8157, "accuracy": 0.59375, "learning_rate": 5.904069343621443e-07, "epoch": 2.3277429783794705, "percentage": 77.67, "elapsed_time": "3:54:24", "remaining_time": "1:07:23"}
+{"current_steps": 1450, "total_steps": 1854, "loss": 0.8274, "accuracy": 0.581250011920929, "learning_rate": 5.633398970942544e-07, "epoch": 2.3439078601737724, "percentage": 78.21, "elapsed_time": "3:55:56", "remaining_time": "1:05:44"}
+{"current_steps": 1460, "total_steps": 1854, "loss": 0.7602, "accuracy": 0.5625, "learning_rate": 5.368289819205069e-07, "epoch": 2.3600727419680743, "percentage": 78.75, "elapsed_time": "3:57:25", "remaining_time": "1:04:04"}
+{"current_steps": 1470, "total_steps": 1854, "loss": 0.9052, "accuracy": 0.5625, "learning_rate": 5.108818015890785e-07, "epoch": 2.376237623762376, "percentage": 79.29, "elapsed_time": "3:59:01", "remaining_time": "1:02:26"}
+{"current_steps": 1480, "total_steps": 1854, "loss": 0.7861, "accuracy": 0.675000011920929, "learning_rate": 4.855058069687291e-07, "epoch": 2.392402505556678, "percentage": 79.83, "elapsed_time": "4:00:31", "remaining_time": "1:00:46"}
+{"current_steps": 1490, "total_steps": 1854, "loss": 0.8935, "accuracy": 0.543749988079071, "learning_rate": 4.607082849092523e-07, "epoch": 2.40856738735098, "percentage": 80.37, "elapsed_time": "4:02:11", "remaining_time": "0:59:09"}
+{"current_steps": 1500, "total_steps": 1854, "loss": 0.8035, "accuracy": 0.5562499761581421, "learning_rate": 4.3649635614901405e-07, "epoch": 2.4247322691452817, "percentage": 80.91, "elapsed_time": "4:03:47", "remaining_time": "0:57:32"}
+{"current_steps": 1500, "total_steps": 1854, "eval_loss": 0.8629826903343201, "epoch": 2.4247322691452817, "percentage": 80.91, "elapsed_time": "4:07:01", "remaining_time": "0:58:17"}