Training in progress, step 6200

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"

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bbb9d05e18d21539814aa33a75bcd59ecfa1423d749f7def1fa099eac295c773
+oid sha256:3febec860772b82c489867736d215315d0909ffc43a3ba7b9fbc1d0b41fd7a20
 size 500897101

{checkpoint-5700 → checkpoint-6000/adapter_model}/adapter_config.json

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

{checkpoint-5800 → checkpoint-6000/adapter_model}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c6c8948b163d7faf80bcb4ea3c791311da3d9b14d4257794aeb4a2b35935a025
+oid sha256:bbb9d05e18d21539814aa33a75bcd59ecfa1423d749f7def1fa099eac295c773
 size 500897101

{checkpoint-5800 → checkpoint-6100}/adapter_config.json

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

{checkpoint-5700 → checkpoint-6100}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6bdd5402d919c62a264000ec96b1dd621956e56ddd66679cf8a429f111552d95
+oid sha256:388cd46150cf53754b4f0d276e98973e5a1f43862ebd435181705f9ab3d56f52
 size 500897101

{checkpoint-5700 → checkpoint-6100}/adapter_model/adapter_config.json

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

{checkpoint-5800 → checkpoint-6100}/adapter_model/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c6c8948b163d7faf80bcb4ea3c791311da3d9b14d4257794aeb4a2b35935a025
+oid sha256:388cd46150cf53754b4f0d276e98973e5a1f43862ebd435181705f9ab3d56f52
 size 500897101

{checkpoint-5700 → checkpoint-6100}/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8847b7875d322ad5570a300587a0f3924a360ae4df3e2feeb97a8a2e21a3cbff
+oid sha256:23361198354be9afc19d0f3752c768fddef1e9604ebeb35127628f0a567c9868
 size 1001752701

{checkpoint-5800 → checkpoint-6100}/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fdcc97c3d5ae4ead7b5285c5c5b8dddcbec730d6aced698514214e40163f6c80
+oid sha256:9162bd2b64fa6a1399b67a40366d8c08f7d3612b17f83dfde4accf31e0b7e38b
 size 27772

{checkpoint-5700 → checkpoint-6100}/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:218ef7d643b85f18451d0ff4e11963ac2483f153fb71cc3ca1ab2bc1c0c67051
+oid sha256:47eb09798665edeada4b80de705ecf394a69e60f01cd9bd340d95b3c9d0be22e
 size 27772

{checkpoint-5800 → checkpoint-6100}/rng_state_10.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:481ee2d6343e3ee249f8f13ba72e97689ec745d34ca3eb731c5977d455e68087
+oid sha256:b9c91cade997cd5e5e47ae09b2a295facdfbba964508a100db5e64faf0320f11
 size 27789

{checkpoint-5800 → checkpoint-6100}/rng_state_11.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:56d4288a9e908f82added428e5a14ff5dd6e86473ebd372733e241f3c2a4e833
+oid sha256:2b313ea64d5b302d9894ee7b5dd63f9c9a469ca1256ceab7619cb9f049f70138
 size 27789

{checkpoint-5700 → checkpoint-6100}/rng_state_12.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8305db4caf39e6fd5e1f06f849d2a6d65681cf5c367dd8a754a886f1d32d4456
+oid sha256:e01b6efa92bd063f4cd80ddd2fa79082d2072f36bc4669abe0ec0742ab75223d
 size 27789

{checkpoint-5800 → checkpoint-6100}/rng_state_13.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2ac848792be0fda751cfdebaa1419f845e441f823e55d97dc332cb6ebecab888
+oid sha256:ab0222a3e50c0985d9dbe1cad74c8abd56400e0d05ff489b51865ef156928695
 size 27789

{checkpoint-5800 → checkpoint-6100}/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:004ca3f2557bec1000ed01ad0ac091380c145ff4d9054e495042d9673b164cd3
+oid sha256:e13e105d97f9b5820cc4ad076585f2902cd3185bac7443faaead5cb3cef83c80
 size 27772

{checkpoint-5800 → checkpoint-6100}/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:83a6e03a2771bdfefb81f6b70a469b27bf46d9bac7c91cec25fbd8fc2e1bb9fd
+oid sha256:8bf0b2a35e4af585c3a853ac8956e69ab3dc44f8f7b289beeed7bc8889382e11
 size 27772

{checkpoint-5700 → checkpoint-6100}/rng_state_4.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:20ebd3abf1496a95cb8a1cc494c974137f7880bb1069b7813817b8600444464d
+oid sha256:daedb950d8b297b35aecd0341c5ab143df51662832bddabd61af69a80c0e2691
 size 27772

{checkpoint-5700 → checkpoint-6100}/rng_state_5.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a0ad3339005efdbfae3bb272ea3ce0907371f54b4834a96fa9c864807063f63d
+oid sha256:debe892fc62955687dbec8db9a770f303801833773975c37decce994212064ee
 size 27772

{checkpoint-5700 → checkpoint-6100}/rng_state_6.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0f13556a5091d1d8f3874fe4a97ac3fc222b586812ceab6167e99bc25dc28fc0
+oid sha256:166b098d25e88227361ec50688e81ef3c0a1c1b007617e16be45748918dd52df
 size 27772

{checkpoint-5700 → checkpoint-6100}/rng_state_7.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6b76a223c2ea9e5262b69fc056b9f29e0dd62a728f58a99ac77374aeb72fce68
+oid sha256:b699249cf37387b69dcd9425364424949a26a56a95817fe5db02f001d3e9c86f
 size 27772

{checkpoint-5700 → checkpoint-6100}/rng_state_8.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ae0f14b82c516564368a80f0d993ef826bf389434bdd326ac982efc3e52dc286
+oid sha256:ba57dff1a1b5008ba90bbde0f66039de21670cc736e37d954752bf67e9cb8130
 size 27772

{checkpoint-5700 → checkpoint-6100}/rng_state_9.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ff589c2de9aa0e8b005f5e548e65245c51b773411b16d01c505e9d491673af3f
+oid sha256:90b54c02081eb6d595752be8c21725bd5098ac7be34d0c63c9a1d26e989ac033
 size 27772

{checkpoint-5700 → checkpoint-6100}/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:935f0139e1e2ffb90e75a6a1904fe50b37854f0d00a6d9b7dc07c85fc7a0d97c
+oid sha256:1bbbf313199bcaa9e78b6da6162b405173b5ef683672f79c9847fdc8ee4406e8
 size 627

{checkpoint-5800 → checkpoint-6100}/trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.2489336952307095,
-  "global_step": 5800,
+  "epoch": 2.365257851880574,
+  "global_step": 6100,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1414,11 +1414,91 @@
       "learning_rate": 2.944041552425475e-05,
       "loss": 0.7687,
       "step": 5800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.8723813270606982e-05,
+      "loss": 0.7698,
+      "step": 5825
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8014574760522416e-05,
+      "loss": 0.7641,
+      "step": 5850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.731277326742876e-05,
+      "loss": 0.7746,
+      "step": 5875
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6618481296414522e-05,
+      "loss": 0.7722,
+      "step": 5900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5931770576738313e-05,
+      "loss": 0.7698,
+      "step": 5925
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.525271205441837e-05,
+      "loss": 0.7751,
+      "step": 5950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.45813758849028e-05,
+      "loss": 0.766,
+      "step": 5975
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3917831425821824e-05,
+      "loss": 0.7673,
+      "step": 6000
+    },
+    {
+      "epoch": 2.33,
+      "eval_loss": 0.806900680065155,
+      "eval_runtime": 58.3254,
+      "eval_samples_per_second": 12.516,
+      "eval_steps_per_second": 0.909,
+      "step": 6000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3262147229821984e-05,
+      "loss": 0.7679,
+      "step": 6025
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2614391037483983e-05,
+      "loss": 0.7704,
+      "step": 6050
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1974629770324106e-05,
+      "loss": 0.7661,
+      "step": 6075
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1342929523880416e-05,
+      "loss": 0.7652,
+      "step": 6100
     }
   ],
   "max_steps": 7737,
   "num_train_epochs": 3,
-  "total_flos": 2.4980935510562177e+19,
+  "total_flos": 2.6271635417086296e+19,
   "trial_name": null,
   "trial_params": null
 }

{checkpoint-5700 → checkpoint-6100}/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:df5d13b3f1b9942f80afde79010ef0947feee3df761d245fef1699bc397648b2
+oid sha256:c008d25bfaf7bb9a08aab93c029ae5c61570d5e3604fc9ca9ea2654e60b13cd6
 size 4027

{checkpoint-5800/adapter_model → checkpoint-6200}/adapter_config.json

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

{checkpoint-5700/adapter_model → checkpoint-6200}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6bdd5402d919c62a264000ec96b1dd621956e56ddd66679cf8a429f111552d95
+oid sha256:3febec860772b82c489867736d215315d0909ffc43a3ba7b9fbc1d0b41fd7a20
 size 500897101

{checkpoint-5800 → checkpoint-6200}/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fdafcf17dcbaf493ce7420b5efc55cc37f121d6d18e479f66229dfb7d3ebfe9a
+oid sha256:c6c238a192e64d590edb62f130ceed462095ecf868445a5da756e160d8783d7e
 size 1001752701

{checkpoint-5700 → checkpoint-6200}/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f06ce8fd6a33ec4407e0888bbcc8c7e40f4d820c9b7b7024f808c1cb08bac1f9
+oid sha256:1a62682ba7e7cb26595ded7f48ec753d2eb9673721db81b0ec4152ba6856f2d2
 size 27772

{checkpoint-5800 → checkpoint-6200}/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8f0dfb5773126adda2e927345797eb8babdfc8aa673a963e413f06bda803f6b8
+oid sha256:74be203c1e4634b2aa1037ec3a4a9edd5a7298915ce425cf1b9a31f6391efca1
 size 27772

{checkpoint-5700 → checkpoint-6200}/rng_state_10.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9d52fed50c25d75d7fa962b9f10f71f848700ff11f9305a37b16ca126fee5481
+oid sha256:c037a2c235445f0da915de6619ff6f7f42b2c0c8575b787c1a4ed16e35b61d21
 size 27789

{checkpoint-5700 → checkpoint-6200}/rng_state_11.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ca4cdcdc956f43b62d0d18dc1d662748eb79bf6714c90145fc90101a1ea2249d
+oid sha256:b62b1f601e03a003f84cf80a8c292113d141ee87f5b1ca845fff2a924a30938e
 size 27789

{checkpoint-5800 → checkpoint-6200}/rng_state_12.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5749fda2bb629d9c3e4a0bb894676f98047fc123c7ada64e2385c1692f194369
+oid sha256:6c758852b4f6179b67806b64c03556a85fc8cb6b2d7fc9b9802e599d92cb5060
 size 27789

{checkpoint-5700 → checkpoint-6200}/rng_state_13.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bd9d3dc3cf0abd594c579dd75cbe79170df6d02cec6c38d47e042b3dd26300d9
+oid sha256:212ebbd04054c53230c6f37560d76104b53022b0398cb77838fa4b24148235e3
 size 27789

{checkpoint-5700 → checkpoint-6200}/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:348612a8cf3a488b2de324430f631e10c4e8156112e6c50cecda44157e829595
+oid sha256:0b20af4a6e33875e1906da2f4f958007730c9409557ca3cdfa5b4dc59f51e2d5
 size 27772

{checkpoint-5700 → checkpoint-6200}/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c688a5f058c380a72ea06519f8b004106c3bf62f4ac8ae5b0e96c473dbac963c
+oid sha256:9f749729546f0e74d9ecf3ad24fea7d29dd1ae76d88741fcdcf02b50df73302c
 size 27772

{checkpoint-5800 → checkpoint-6200}/rng_state_4.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:027cc230cd1577ece8a43abc1ed88fcc51ac874f84b8bcbe865b8338b17825dc
+oid sha256:0c21e5a49b0e24cc974d1d7d4406b9728cbea5332e18346ba9afb35e0372f526
 size 27772

{checkpoint-5800 → checkpoint-6200}/rng_state_5.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5e15c6848b49b0375761f21215e2d56ea5a16e27885192cc4af7b1a3f9c325a2
+oid sha256:3a33c776cc1ad3767cdac6f06b81c8d34fcc1d3ad73a9f8803254d0d817aa09a
 size 27772

{checkpoint-5800 → checkpoint-6200}/rng_state_6.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0290c211df0a25ca2b739876f096befd7a2f50010f23fcfb700c648fc7c37b88
+oid sha256:64fcf9ef649f08b421dd4acc8e2b02fc8f3ffefc82b05ef17dde8609c9d4bb40
 size 27772

{checkpoint-5800 → checkpoint-6200}/rng_state_7.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fe7bdf51b5f2867802a2660bdbfa211be18d28a9f5e7325d4957f7ee65895a3a
+oid sha256:cbea4f476c6f296b74387728b4c7179586b903ae8344e377cc5ce9725f14ba69
 size 27772

{checkpoint-5800 → checkpoint-6200}/rng_state_8.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0c7043986c3cfa7014abbd2eb87d4f8ccda04f6a243deda922c9522e5c66f2fd
+oid sha256:ecac165f822fec78a8825820ff7d48e36aa3f53cb41f7a57d2e9dfadf8de351a
 size 27772

{checkpoint-5800 → checkpoint-6200}/rng_state_9.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a3371033bac6e8e8c6fb2001ca45e33bd941b91fd370cce2d2edc6bac7df0e55
+oid sha256:ba8091a98c98de998b6ead9067c77d9531ff0bb7b308e548615a3a1c2ec0f7e8
 size 27772

{checkpoint-5800 → checkpoint-6200}/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d7e2a15463250851f8f0353fd93191418e2b4e589eb68be6185d12012957a33f
+oid sha256:1a2465347a031d34e4adae1fd19f3f075689a8b84714f2008d365cd61979bafe
 size 627

{checkpoint-5700 → checkpoint-6200}/trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.2101589763474214,
-  "global_step": 5700,
+  "epoch": 2.404032570763862,
+  "global_step": 6200,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1390,11 +1390,139 @@
       "learning_rate": 3.237897004017276e-05,
       "loss": 0.7716,
       "step": 5700
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.163366064673427e-05,
+      "loss": 0.7721,
+      "step": 5725
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.089541437233252e-05,
+      "loss": 0.7658,
+      "step": 5750
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0164307487268996e-05,
+      "loss": 0.7716,
+      "step": 5775
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.944041552425475e-05,
+      "loss": 0.7687,
+      "step": 5800
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.8723813270606982e-05,
+      "loss": 0.7698,
+      "step": 5825
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.8014574760522416e-05,
+      "loss": 0.7641,
+      "step": 5850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.731277326742876e-05,
+      "loss": 0.7746,
+      "step": 5875
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6618481296414522e-05,
+      "loss": 0.7722,
+      "step": 5900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5931770576738313e-05,
+      "loss": 0.7698,
+      "step": 5925
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.525271205441837e-05,
+      "loss": 0.7751,
+      "step": 5950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.45813758849028e-05,
+      "loss": 0.766,
+      "step": 5975
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3917831425821824e-05,
+      "loss": 0.7673,
+      "step": 6000
+    },
+    {
+      "epoch": 2.33,
+      "eval_loss": 0.806900680065155,
+      "eval_runtime": 58.3254,
+      "eval_samples_per_second": 12.516,
+      "eval_steps_per_second": 0.909,
+      "step": 6000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3262147229821984e-05,
+      "loss": 0.7679,
+      "step": 6025
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2614391037483983e-05,
+      "loss": 0.7704,
+      "step": 6050
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1974629770324106e-05,
+      "loss": 0.7661,
+      "step": 6075
+    },
+    {
+      "epoch": 2.37,
+      "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
     }
   ],
   "max_steps": 7737,
   "num_train_epochs": 3,
-  "total_flos": 2.4549667828507083e+19,
+  "total_flos": 2.6702654094117437e+19,
   "trial_name": null,
   "trial_params": null
 }

{checkpoint-5800 → checkpoint-6200}/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:df5d13b3f1b9942f80afde79010ef0947feee3df761d245fef1699bc397648b2
+oid sha256:e85009bca9623c846e630c294adb80ecbcd9e720da8da9f9ee5311b562908b91
 size 4027