Training in progress, step 6600

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eca317a26dab233b92de0d6d401d39cb8b1fe17ca51e4a191e73b88cb76bfc3d
+oid sha256:d81f9de68995df3ae7fe51e14a4d5a0482b3ff50173f115d320ee99247cb5e3d
 size 500897101

{checkpoint-6200 → checkpoint-6500/adapter_model}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3febec860772b82c489867736d215315d0909ffc43a3ba7b9fbc1d0b41fd7a20
+oid sha256:eca317a26dab233b92de0d6d401d39cb8b1fe17ca51e4a191e73b88cb76bfc3d
 size 500897101

{checkpoint-6200/adapter_model → checkpoint-6600}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3febec860772b82c489867736d215315d0909ffc43a3ba7b9fbc1d0b41fd7a20
+oid sha256:d81f9de68995df3ae7fe51e14a4d5a0482b3ff50173f115d320ee99247cb5e3d
 size 500897101

{checkpoint-6200 → checkpoint-6600}/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c6c238a192e64d590edb62f130ceed462095ecf868445a5da756e160d8783d7e
+oid sha256:126ec0322a5714bfa1d9503fe376d8ebf71cc64ffc913454c2c8b48b06bfd15f
 size 1001752701

{checkpoint-6200 → checkpoint-6600}/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1a62682ba7e7cb26595ded7f48ec753d2eb9673721db81b0ec4152ba6856f2d2
+oid sha256:6a6a609c8ce73fc706f6fb966ebfd028a35b6fc2b3ef2c3b90b14c2d9d96c073
 size 27772

{checkpoint-6200 → checkpoint-6600}/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:74be203c1e4634b2aa1037ec3a4a9edd5a7298915ce425cf1b9a31f6391efca1
+oid sha256:513344a4102cc8801ca6a9111665c65ff7e9a337b61788efc4dbd10afd40421e
 size 27772

{checkpoint-6200 → checkpoint-6600}/rng_state_10.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c037a2c235445f0da915de6619ff6f7f42b2c0c8575b787c1a4ed16e35b61d21
+oid sha256:84bdaf5f54f27ae7dd4367ad1df5dc5c81c793f62be76df32dd2a92891c0ae9f
 size 27789

{checkpoint-6200 → checkpoint-6600}/rng_state_11.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b62b1f601e03a003f84cf80a8c292113d141ee87f5b1ca845fff2a924a30938e
+oid sha256:98e58aa58921c7b94e7c72eb60e9c4d6abe4187fa10dfd3dde14df9291075366
 size 27789

{checkpoint-6200 → checkpoint-6600}/rng_state_12.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6c758852b4f6179b67806b64c03556a85fc8cb6b2d7fc9b9802e599d92cb5060
+oid sha256:90a0a154457600ec617c2bdbb216fa73b29edafbe85cd1f69bdc45d06510b354
 size 27789

{checkpoint-6200 → checkpoint-6600}/rng_state_13.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:212ebbd04054c53230c6f37560d76104b53022b0398cb77838fa4b24148235e3
+oid sha256:702356496f60d3e621103bc084d4848106ba4006fdd2d5ad6640539693a5e42a
 size 27789

{checkpoint-6200 → checkpoint-6600}/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0b20af4a6e33875e1906da2f4f958007730c9409557ca3cdfa5b4dc59f51e2d5
+oid sha256:046a0af63804ca31a141d5015a52b23ee2774bb6e4992759b32c342ddc173129
 size 27772

{checkpoint-6200 → checkpoint-6600}/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9f749729546f0e74d9ecf3ad24fea7d29dd1ae76d88741fcdcf02b50df73302c
+oid sha256:4eb3a5e7cd3e2477ec830ebb148f6b3b4ae734efaa4086c861d75bd02f98d32d
 size 27772

{checkpoint-6200 → checkpoint-6600}/rng_state_4.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0c21e5a49b0e24cc974d1d7d4406b9728cbea5332e18346ba9afb35e0372f526
+oid sha256:8dfd38532c9714d2d07dcbded515916fc4a2a8ebf7809eb89c8e40ba30623300
 size 27772

{checkpoint-6200 → checkpoint-6600}/rng_state_5.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3a33c776cc1ad3767cdac6f06b81c8d34fcc1d3ad73a9f8803254d0d817aa09a
+oid sha256:f6d24543c66c5d1f88ce0e125b12e7a3822599ea0fc5c8774770d4b6e36e9a0a
 size 27772

{checkpoint-6200 → checkpoint-6600}/rng_state_6.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:64fcf9ef649f08b421dd4acc8e2b02fc8f3ffefc82b05ef17dde8609c9d4bb40
+oid sha256:e194294db8d8734e6cc4d219d1f312dace676b799dac00f96ab1b32058105929
 size 27772

{checkpoint-6200 → checkpoint-6600}/rng_state_7.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cbea4f476c6f296b74387728b4c7179586b903ae8344e377cc5ce9725f14ba69
+oid sha256:1970a2d9f7cb4f5fff840955a3b522a1b4bb096917d81f71d05d1ab7512a2d1a
 size 27772

{checkpoint-6200 → checkpoint-6600}/rng_state_8.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ecac165f822fec78a8825820ff7d48e36aa3f53cb41f7a57d2e9dfadf8de351a
+oid sha256:e4d396595cc95649a6e4b094f41829540f57ac9e24ad4e6eb37d58aac423d45a
 size 27772

{checkpoint-6200 → checkpoint-6600}/rng_state_9.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ba8091a98c98de998b6ead9067c77d9531ff0bb7b308e548615a3a1c2ec0f7e8
+oid sha256:aed28956ad5859c504e2fb1745186ccc74f16e74b31582a197cb9dcab20b3d56
 size 27772

{checkpoint-6200 → checkpoint-6600}/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1a2465347a031d34e4adae1fd19f3f075689a8b84714f2008d365cd61979bafe
+oid sha256:152f426be01aff254cb20a567c7989c603706f46efd9a2a941b2d4911bff72ad
 size 627

{checkpoint-6200 → checkpoint-6600}/trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.404032570763862,
-  "global_step": 6200,
+  "epoch": 2.5591314462970143,
+  "global_step": 6600,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1518,11 +1518,107 @@
       "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
+    },
+    {
+      "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
     }
   ],
   "max_steps": 7737,
   "num_train_epochs": 3,
-  "total_flos": 2.6702654094117437e+19,
+  "total_flos": 2.842466784500515e+19,
   "trial_name": null,
   "trial_params": null
 }