Training in progress, step 6900

adapter_config.json

@@ -14,13 +14,13 @@
   "r": 32,
   "revision": null,
   "target_modules": [
-    "up_proj",
-    "gate_proj",
     "down_proj",
+    "up_proj",
     "q_proj",
     "k_proj",
-    "o_proj",
-    "v_proj"
+    "gate_proj",
+    "v_proj",
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:caee01e1da9f7e92223499414f8bb25b28ea34daf6fb927c9bd869dcba2559fd
+oid sha256:6dce28954f69ce7f466b50fc774c94fb6da869b76f94baf208c3e276b1365db2
 size 500897101

{checkpoint-6500 → checkpoint-6800/adapter_model}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eca317a26dab233b92de0d6d401d39cb8b1fe17ca51e4a191e73b88cb76bfc3d
+oid sha256:caee01e1da9f7e92223499414f8bb25b28ea34daf6fb927c9bd869dcba2559fd
 size 500897101

{checkpoint-6500/adapter_model → checkpoint-6900}/adapter_config.json

@@ -14,13 +14,13 @@
   "r": 32,
   "revision": null,
   "target_modules": [
-    "up_proj",
-    "gate_proj",
     "down_proj",
+    "up_proj",
     "q_proj",
     "k_proj",
-    "o_proj",
-    "v_proj"
+    "gate_proj",
+    "v_proj",
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM"
 }

{checkpoint-6500/adapter_model → checkpoint-6900}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eca317a26dab233b92de0d6d401d39cb8b1fe17ca51e4a191e73b88cb76bfc3d
+oid sha256:6dce28954f69ce7f466b50fc774c94fb6da869b76f94baf208c3e276b1365db2
 size 500897101

{checkpoint-6500 → checkpoint-6900}/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d9682eef3ae7298917bb748df54df8acc39f0813218a1379d7f07289466ceb1f
+oid sha256:5e51694cb0a1cd9603ece0232dabcd7a370a6f8e378f0b01213ceaaf413c4d0c
 size 1001752701

{checkpoint-6500 → checkpoint-6900}/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dbf0d11080193db9fc57ef80bb7e4db0c2dfdb81c6c7d949934ecea2b60c19bf
+oid sha256:361a0cbcd4b75d795f0109f290c642aca880cc06772f00227ba1e0217fc939ef
 size 27772

{checkpoint-6500 → checkpoint-6900}/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:55ed8dac28e473953d59209a3d2e84f582e221e1aeec01bb889b689dd5d44e2d
+oid sha256:0f5123ae610aa26457a3fa34fd70ab7e4ab44cb21dc09fd095aad55a2d15294a
 size 27772

{checkpoint-6500 → checkpoint-6900}/rng_state_10.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ccf2c7c92e70421594d6f3f46563c37fcacbab1b490cdc37991945fbefd0f763
+oid sha256:a213c64ebe2689a3f02bbe5b0ee2c66feb57c0f5f4f8621c0ec2f63c6c590892
 size 27789

{checkpoint-6500 → checkpoint-6900}/rng_state_11.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b57753b4305e0903d863e858d017bc2028a3f7434a2c1e604d81db6a5380681d
+oid sha256:531d6f0912536de9c8ce68b83c17fc005458796d7921c1fe083548fda04d57a7
 size 27789

{checkpoint-6500 → checkpoint-6900}/rng_state_12.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e3a73c535e411cb5175cf8662e504f0b075dc27417beaf98fb6764a53d08c703
+oid sha256:dad9e94de76a75ad267b7650de4df602944013ed9b553b148fa94de6ea86bb52
 size 27789

{checkpoint-6500 → checkpoint-6900}/rng_state_13.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8c48a0a908e8938ff0202d305b73a93eef0a0f00a8215a6dc22758d9da73ca3d
+oid sha256:7f9e0a77144d5a662958f707fc8b494554054fd279f5a3dc74159dc97e6d1e78
 size 27789

{checkpoint-6500 → checkpoint-6900}/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:25834dbb3472e6bffcb9ff8e2e1078b3d7285d899c32d530801c909e58792ca0
+oid sha256:c5b931b3c81ed25f1fcd0971b6da08a3869bfd28a0489528ed3c96c3619fa557
 size 27772

{checkpoint-6500 → checkpoint-6900}/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6f2243c59fbd105663e2bc1078c5d294f026256157124a120dd6fc96481dde2a
+oid sha256:2e121c889141260915892ddfe435284a35a9f246ef67720bdc33081a846ab328
 size 27772

{checkpoint-6500 → checkpoint-6900}/rng_state_4.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:80e34c823c057e9fa9761cbcb6a98def6aedfc20291fc6e93777c2fd857362da
+oid sha256:3f912038cfe5be27104676eb3640c3beb52c1c4c80bafea1732f0443a67ad4b4
 size 27772

{checkpoint-6500 → checkpoint-6900}/rng_state_5.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:49a370bb0d4117c8cf565138b86c7c35fdd0f6916cd6a676aff48559b5050a97
+oid sha256:7e7e05a90589de96c2e2748fa928890eb3ae0f6585d2b9768316f612aa8cf908
 size 27772

{checkpoint-6500 → checkpoint-6900}/rng_state_6.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a142e199df177be102679c7c5299cd0ca74e71330003c3b0c22bf166a55fd738
+oid sha256:dfaec3c28cf52d29cdacae2acf43363d9aea5a881877b06a0055a14c1b7aed6a
 size 27772

{checkpoint-6500 → checkpoint-6900}/rng_state_7.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fdd64c508edb14ee20f09cd5f53a1d6ae1b6ac188a5ce19c88b43bb1c51e651c
+oid sha256:c35c98f247077b0ff5a006aeba6561c7e6bf2e9fd8969c9bc35d64aaf01cf14a
 size 27772

{checkpoint-6500 → checkpoint-6900}/rng_state_8.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:117885c5fa979f628afc9a7003d4946141b0153b62ce3ffa6875857ffdd78ae8
+oid sha256:86a908ac53361e9d5c34680f8e1789e69824f75a3e0b100140c8133e52f8fbed
 size 27772

{checkpoint-6500 → checkpoint-6900}/rng_state_9.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9794ac48303e5da2f12696c803580c6cec30009df7c1a5151886c1bb7932c058
+oid sha256:5597a151c26d0f9f0ba329a3cf67af7f95b7fed3d58ea4461f586948ea11f890
 size 27772

{checkpoint-6500 → checkpoint-6900}/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5c2bdf46b70bc586fca89e010bc74dca67e64afb8ffe6379ceecc7b272fb7824
+oid sha256:ed903a2193f395f38212602a53ab2a5c251266f0ae7d7dc0557e3b565ed8e240
 size 627

{checkpoint-6500 → checkpoint-6900}/trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.520356727413726,
-  "global_step": 6500,
+  "epoch": 2.6754556029468786,
+  "global_step": 6900,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1590,11 +1590,107 @@
       "learning_rate": 1.2382654384049475e-05,
       "loss": 0.7714,
       "step": 6500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.1897272010289884e-05,
+      "loss": 0.7701,
+      "step": 6525
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1420991777454315e-05,
+      "loss": 0.7628,
+      "step": 6550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0953862891394795e-05,
+      "loss": 0.7661,
+      "step": 6575
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0495933612511976e-05,
+      "loss": 0.7729,
+      "step": 6600
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0047251250769175e-05,
+      "loss": 0.772,
+      "step": 6625
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.60786216080466e-06,
+      "loss": 0.7702,
+      "step": 6650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.177811737142627e-06,
+      "loss": 0.7711,
+      "step": 6675
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.757144409503359e-06,
+      "loss": 0.765,
+      "step": 6700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.34590363821306e-06,
+      "loss": 0.7713,
+      "step": 6725
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.944131909713859e-06,
+      "loss": 0.7631,
+      "step": 6750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.551870732174416e-06,
+      "loss": 0.767,
+      "step": 6775
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.169160631201566e-06,
+      "loss": 0.7692,
+      "step": 6800
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.796041145653553e-06,
+      "loss": 0.7677,
+      "step": 6825
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.432550823555128e-06,
+      "loss": 0.7706,
+      "step": 6850
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.078727218115043e-06,
+      "loss": 0.7678,
+      "step": 6875
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.734606883846338e-06,
+      "loss": 0.7717,
+      "step": 6900
     }
   ],
   "max_steps": 7737,
   "num_train_epochs": 3,
-  "total_flos": 2.7994166247676707e+19,
+  "total_flos": 2.9715299616168083e+19,
   "trial_name": null,
   "trial_params": null
 }

{checkpoint-6500 → checkpoint-6900}/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e85009bca9623c846e630c294adb80ecbcd9e720da8da9f9ee5311b562908b91
+oid sha256:1c2f4cc723c65538de12d445f71a30b1610b702bfe771edb2385636bb8724bfd
 size 4027

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e85009bca9623c846e630c294adb80ecbcd9e720da8da9f9ee5311b562908b91
+oid sha256:1c2f4cc723c65538de12d445f71a30b1610b702bfe771edb2385636bb8724bfd
 size 4027