Training in progress, step 6500

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:90226543ac07d2ffe305a17c9351436408a800fb2798a5a37995d29cfa11cdec
+oid sha256:eca317a26dab233b92de0d6d401d39cb8b1fe17ca51e4a191e73b88cb76bfc3d
 size 500897101

{checkpoint-6100 → checkpoint-6400/adapter_model}/adapter_config.json

@@ -14,12 +14,12 @@
   "r": 32,
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "down_proj",
-    "o_proj",
     "up_proj",
     "gate_proj",
+    "down_proj",
     "q_proj",
+    "k_proj",
+    "o_proj",
     "v_proj"
   ],
   "task_type": "CAUSAL_LM"

{checkpoint-6100 → checkpoint-6400/adapter_model}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:388cd46150cf53754b4f0d276e98973e5a1f43862ebd435181705f9ab3d56f52
+oid sha256:90226543ac07d2ffe305a17c9351436408a800fb2798a5a37995d29cfa11cdec
 size 500897101

{checkpoint-6100/adapter_model → checkpoint-6500}/adapter_config.json

@@ -14,12 +14,12 @@
   "r": 32,
   "revision": null,
   "target_modules": [
-    "k_proj",
-    "down_proj",
-    "o_proj",
     "up_proj",
     "gate_proj",
+    "down_proj",
     "q_proj",
+    "k_proj",
+    "o_proj",
     "v_proj"
   ],
   "task_type": "CAUSAL_LM"

{checkpoint-6100/adapter_model → checkpoint-6500}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:388cd46150cf53754b4f0d276e98973e5a1f43862ebd435181705f9ab3d56f52
+oid sha256:eca317a26dab233b92de0d6d401d39cb8b1fe17ca51e4a191e73b88cb76bfc3d
 size 500897101

{checkpoint-6100 → checkpoint-6500}/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:23361198354be9afc19d0f3752c768fddef1e9604ebeb35127628f0a567c9868
+oid sha256:d9682eef3ae7298917bb748df54df8acc39f0813218a1379d7f07289466ceb1f
 size 1001752701

{checkpoint-6100 → checkpoint-6500}/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9162bd2b64fa6a1399b67a40366d8c08f7d3612b17f83dfde4accf31e0b7e38b
+oid sha256:dbf0d11080193db9fc57ef80bb7e4db0c2dfdb81c6c7d949934ecea2b60c19bf
 size 27772

{checkpoint-6100 → checkpoint-6500}/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:47eb09798665edeada4b80de705ecf394a69e60f01cd9bd340d95b3c9d0be22e
+oid sha256:55ed8dac28e473953d59209a3d2e84f582e221e1aeec01bb889b689dd5d44e2d
 size 27772

{checkpoint-6100 → checkpoint-6500}/rng_state_10.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b9c91cade997cd5e5e47ae09b2a295facdfbba964508a100db5e64faf0320f11
+oid sha256:ccf2c7c92e70421594d6f3f46563c37fcacbab1b490cdc37991945fbefd0f763
 size 27789

{checkpoint-6100 → checkpoint-6500}/rng_state_11.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2b313ea64d5b302d9894ee7b5dd63f9c9a469ca1256ceab7619cb9f049f70138
+oid sha256:b57753b4305e0903d863e858d017bc2028a3f7434a2c1e604d81db6a5380681d
 size 27789

{checkpoint-6100 → checkpoint-6500}/rng_state_12.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e01b6efa92bd063f4cd80ddd2fa79082d2072f36bc4669abe0ec0742ab75223d
+oid sha256:e3a73c535e411cb5175cf8662e504f0b075dc27417beaf98fb6764a53d08c703
 size 27789

{checkpoint-6100 → checkpoint-6500}/rng_state_13.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ab0222a3e50c0985d9dbe1cad74c8abd56400e0d05ff489b51865ef156928695
+oid sha256:8c48a0a908e8938ff0202d305b73a93eef0a0f00a8215a6dc22758d9da73ca3d
 size 27789

{checkpoint-6100 → checkpoint-6500}/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e13e105d97f9b5820cc4ad076585f2902cd3185bac7443faaead5cb3cef83c80
+oid sha256:25834dbb3472e6bffcb9ff8e2e1078b3d7285d899c32d530801c909e58792ca0
 size 27772

{checkpoint-6100 → checkpoint-6500}/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8bf0b2a35e4af585c3a853ac8956e69ab3dc44f8f7b289beeed7bc8889382e11
+oid sha256:6f2243c59fbd105663e2bc1078c5d294f026256157124a120dd6fc96481dde2a
 size 27772

{checkpoint-6100 → checkpoint-6500}/rng_state_4.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:daedb950d8b297b35aecd0341c5ab143df51662832bddabd61af69a80c0e2691
+oid sha256:80e34c823c057e9fa9761cbcb6a98def6aedfc20291fc6e93777c2fd857362da
 size 27772

{checkpoint-6100 → checkpoint-6500}/rng_state_5.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:debe892fc62955687dbec8db9a770f303801833773975c37decce994212064ee
+oid sha256:49a370bb0d4117c8cf565138b86c7c35fdd0f6916cd6a676aff48559b5050a97
 size 27772

{checkpoint-6100 → checkpoint-6500}/rng_state_6.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:166b098d25e88227361ec50688e81ef3c0a1c1b007617e16be45748918dd52df
+oid sha256:a142e199df177be102679c7c5299cd0ca74e71330003c3b0c22bf166a55fd738
 size 27772

{checkpoint-6100 → checkpoint-6500}/rng_state_7.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b699249cf37387b69dcd9425364424949a26a56a95817fe5db02f001d3e9c86f
+oid sha256:fdd64c508edb14ee20f09cd5f53a1d6ae1b6ac188a5ce19c88b43bb1c51e651c
 size 27772

{checkpoint-6100 → checkpoint-6500}/rng_state_8.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ba57dff1a1b5008ba90bbde0f66039de21670cc736e37d954752bf67e9cb8130
+oid sha256:117885c5fa979f628afc9a7003d4946141b0153b62ce3ffa6875857ffdd78ae8
 size 27772

{checkpoint-6100 → checkpoint-6500}/rng_state_9.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:90b54c02081eb6d595752be8c21725bd5098ac7be34d0c63c9a1d26e989ac033
+oid sha256:9794ac48303e5da2f12696c803580c6cec30009df7c1a5151886c1bb7932c058
 size 27772

{checkpoint-6100 → checkpoint-6500}/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1bbbf313199bcaa9e78b6da6162b405173b5ef683672f79c9847fdc8ee4406e8
+oid sha256:5c2bdf46b70bc586fca89e010bc74dca67e64afb8ffe6379ceecc7b272fb7824
 size 627

{checkpoint-6100 → checkpoint-6500}/trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.365257851880574,
-  "global_step": 6100,
+  "epoch": 2.520356727413726,
+  "global_step": 6500,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1494,11 +1494,107 @@
       "learning_rate": 2.1342929523880416e-05,
       "loss": 0.7652,
       "step": 6100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0719355560884246e-05,
+      "loss": 0.765,
+      "step": 6125
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.010397230451766e-05,
+      "loss": 0.7704,
+      "step": 6150
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9496843331757784e-05,
+      "loss": 0.767,
+      "step": 6175
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8898031366808467e-05,
+      "loss": 0.7654,
+      "step": 6200
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.830759827462004e-05,
+      "loss": 0.7753,
+      "step": 6225
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7725605054497906e-05,
+      "loss": 0.7725,
+      "step": 6250
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7152111833800522e-05,
+      "loss": 0.7698,
+      "step": 6275
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6587177861727454e-05,
+      "loss": 0.7703,
+      "step": 6300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6030861503198204e-05,
+      "loss": 0.7658,
+      "step": 6325
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.548322023282235e-05,
+      "loss": 0.7677,
+      "step": 6350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.4944310628961666e-05,
+      "loss": 0.7694,
+      "step": 6375
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4414188367884907e-05,
+      "loss": 0.7668,
+      "step": 6400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3892908218015654e-05,
+      "loss": 0.7662,
+      "step": 6425
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3380524034274122e-05,
+      "loss": 0.7689,
+      "step": 6450
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.2877088752513189e-05,
+      "loss": 0.7694,
+      "step": 6475
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2382654384049475e-05,
+      "loss": 0.7714,
+      "step": 6500
     }
   ],
   "max_steps": 7737,
   "num_train_epochs": 3,
-  "total_flos": 2.6271635417086296e+19,
+  "total_flos": 2.7994166247676707e+19,
   "trial_name": null,
   "trial_params": null
 }

{checkpoint-6100 → checkpoint-6500}/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c008d25bfaf7bb9a08aab93c029ae5c61570d5e3604fc9ca9ea2654e60b13cd6
+oid sha256:e85009bca9623c846e630c294adb80ecbcd9e720da8da9f9ee5311b562908b91
 size 4027