Training in progress, step 4900

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ff09dc21b35d5903035d753d146589f20bd9535c0d4b5f1426eb089c6ad99842
+oid sha256:2a3bb33a0c4878b032aec2026e3879865fa62989d9939826a716eb3580fa34cd
 size 500897101

{checkpoint-4500 → checkpoint-4800/adapter_model}/adapter_config.json

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

{checkpoint-4500 → checkpoint-4800/adapter_model}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b86105f8dc5ea0becf68a83254e59fe9b025467416108e48782683b1c607433c
+oid sha256:ff09dc21b35d5903035d753d146589f20bd9535c0d4b5f1426eb089c6ad99842
 size 500897101

{checkpoint-4500/adapter_model → checkpoint-4900}/adapter_config.json

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

{checkpoint-4500/adapter_model → checkpoint-4900}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b86105f8dc5ea0becf68a83254e59fe9b025467416108e48782683b1c607433c
+oid sha256:2a3bb33a0c4878b032aec2026e3879865fa62989d9939826a716eb3580fa34cd
 size 500897101

{checkpoint-4500 → checkpoint-4900}/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:85174da2f89a11129da995afaf6255c91a1d5b6e403d37872dba3a3b3a40e645
+oid sha256:89baee7be101e0977ca1b6a39db7917193704b00b2bc0dc4bbf1b2e60e7e6a22
 size 1001752701

{checkpoint-4500 → checkpoint-4900}/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a5bc782c5282905b651f7c3d9350387b4d4215bf5d50b4b2a5b37bf164e97dc7
+oid sha256:3d2fb5eb561d4b92c89062c2939861ac27d199067133d2975e1e2aecd6254425
 size 27772

{checkpoint-4500 → checkpoint-4900}/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:11484a14c79259042640491741e6f37434dad1461bbd58f5232d17b08b8e533e
+oid sha256:845f7f7308d4eb8331386ad93aebcacf5679b5a7f2b26d79e0e881c33768d801
 size 27772

{checkpoint-4500 → checkpoint-4900}/rng_state_10.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f42f6efc0683a049d6baa7df6a7c0e43d97448fcc55769f46cad26a5531017ab
+oid sha256:ac19165b3d0d8492404b2597a692a25174dc25bbebee0818cc70fb91e7fa66a9
 size 27789

{checkpoint-4500 → checkpoint-4900}/rng_state_11.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0fcfaf931e665e34bf45012b6b6a1c63046e3744f5af54e03d4f96c348e18585
+oid sha256:8ba175f65490c813c43be1b8e90c05273b365a5fc35034a39d3e40171b6649d8
 size 27789

{checkpoint-4500 → checkpoint-4900}/rng_state_12.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eafb561ee7e481b6d7114e7bd288bbf11e3bdb5b30ccf7671abf23d395f85099
+oid sha256:ff1e6d6959816301feff7a9806d0ca028fe2a211e238b9dca8fd390289758baa
 size 27789

{checkpoint-4500 → checkpoint-4900}/rng_state_13.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:21ed99a2eab0b2ac12b12649669542073aea799df1338d4a51bd73cdf6b630ee
+oid sha256:f1211e56e1d3e784cd9f700dcd2e4b8d94627474333f9ea95573cbff916f6590
 size 27789

{checkpoint-4500 → checkpoint-4900}/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3ed4fa478718a75f5b8e4e8335dccc7f8a0c6fe0000bf381eab0bed5cc01452f
+oid sha256:1a4a16265fbd33a987004f285a9ac8e9d11fa74e2f1e1ebfae43b07d355feb1a
 size 27772

{checkpoint-4500 → checkpoint-4900}/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fe5db2a4b8b4463e754a38ab1d770b8505eab9cf9237cd9ca02883ae04de9553
+oid sha256:25be8bdc125a49f8faa5491285fe8c51cbd061cd60a12c31e1d294969eaeed1b
 size 27772

{checkpoint-4500 → checkpoint-4900}/rng_state_4.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6593b14ae61ea6f6cac1403a93aa234ce2bed31bdab3012398697af88403060a
+oid sha256:f8edc43261484ceb88153a0cf622a2f1d79ae2bd2a22d2cfc258fbc4d67c9fcf
 size 27772

{checkpoint-4500 → checkpoint-4900}/rng_state_5.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6671c2d456854ff131f2e4cc9390c08e9d64ea65a071a9e2fb31165c3c3954bc
+oid sha256:f3c01adf24a8fb5ced49e768a00799034bf23fdbe170faf373d7bffc71cf330e
 size 27772

{checkpoint-4500 → checkpoint-4900}/rng_state_6.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c6a25a453bf0407fdcd7e6a37805f49d6fde89bc6edcd280f665246b8154f7aa
+oid sha256:1431fdce90c9dffa41af7e45225d7e937f8a892f36234dd2c9c0c5ea825d344b
 size 27772

{checkpoint-4500 → checkpoint-4900}/rng_state_7.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8fbd00e43453f6e3e6dc5ec92264277cd945e007b31fe87eadf6722d8a9630a7
+oid sha256:698169f9adbf09c9e392876211e57228599d852964332781ccccb91b02c99668
 size 27772

{checkpoint-4500 → checkpoint-4900}/rng_state_8.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c2261744a6ab0c7c3113ef9956e2adc056db650271f411ba606d73cbea19559f
+oid sha256:1e12b62affc946cb152ae7675f1074ed4a8c28a930020230d21a02278a8cfc66
 size 27772

{checkpoint-4500 → checkpoint-4900}/rng_state_9.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3ec26978e66063d1acde7f517a82236a206de3c0d685eedcf64e8ed048952611
+oid sha256:8d0857929e84e3ee4f8ead280fd3f99826ceb6e6e749cce250845a12d23d0cc7
 size 27772

{checkpoint-4500 → checkpoint-4900}/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ae7fcacd784ea944eefe300a2ada2a5c60a0313b9b263bf62a3a0f9ef8cdef55
+oid sha256:fce2b4a13969f53fdf68394680e1b829aa39975e2753c73d5d093518e2315b1c
 size 627

{checkpoint-4500 → checkpoint-4900}/trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.7448623497479643,
-  "global_step": 4500,
+  "epoch": 1.8999612252811167,
+  "global_step": 4900,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1094,11 +1094,107 @@
       "learning_rate": 7.480272555710227e-05,
       "loss": 0.8006,
       "step": 4500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.382040650400185e-05,
+      "loss": 0.7937,
+      "step": 4525
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.28407921384267e-05,
+      "loss": 0.794,
+      "step": 4550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.186398366709545e-05,
+      "loss": 0.7931,
+      "step": 4575
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.089008200684197e-05,
+      "loss": 0.7982,
+      "step": 4600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.991918777418928e-05,
+      "loss": 0.7916,
+      "step": 4625
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.895140127495455e-05,
+      "loss": 0.7919,
+      "step": 4650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.798682249388631e-05,
+      "loss": 0.7863,
+      "step": 4675
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.702555108433461e-05,
+      "loss": 0.789,
+      "step": 4700
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.606768635795574e-05,
+      "loss": 0.7902,
+      "step": 4725
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.511332727445191e-05,
+      "loss": 0.7924,
+      "step": 4750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.416257243134747e-05,
+      "loss": 0.7957,
+      "step": 4775
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.321552005380256e-05,
+      "loss": 0.7916,
+      "step": 4800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.22722679844652e-05,
+      "loss": 0.7867,
+      "step": 4825
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.133291367336284e-05,
+      "loss": 0.7944,
+      "step": 4850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.039755416783457e-05,
+      "loss": 0.7982,
+      "step": 4875
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.946628610250484e-05,
+      "loss": 0.7918,
+      "step": 4900
     }
   ],
   "max_steps": 7737,
   "num_train_epochs": 3,
-  "total_flos": 1.937872481556837e+19,
+  "total_flos": 2.1101658419601867e+19,
   "trial_name": null,
   "trial_params": null
 }

{checkpoint-4500 → checkpoint-4900}/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4458bf553277b063d908ed9668b95abc04892c52d8d793f6b007433394d06f17
+oid sha256:354c4769acaf21317d8ab58ba2dd133763f3bee1aec7488fb334e45f3e80ebc3
 size 4027