Training in progress, step 4600

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b86105f8dc5ea0becf68a83254e59fe9b025467416108e48782683b1c607433c
+oid sha256:68baca464c45ad2c5da146b800b15dadf06eb685f6905c3cba44f31d2963c595
 size 500897101

{checkpoint-4200 → checkpoint-4500/adapter_model}/adapter_config.json

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

{checkpoint-4200 → checkpoint-4500/adapter_model}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f3e333df20d090244c536ca681f370106e84c45d6f6033e42293d5dd935148eb
+oid sha256:b86105f8dc5ea0becf68a83254e59fe9b025467416108e48782683b1c607433c
 size 500897101

{checkpoint-4200/adapter_model → checkpoint-4600}/adapter_config.json

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

{checkpoint-4200/adapter_model → checkpoint-4600}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f3e333df20d090244c536ca681f370106e84c45d6f6033e42293d5dd935148eb
+oid sha256:68baca464c45ad2c5da146b800b15dadf06eb685f6905c3cba44f31d2963c595
 size 500897101

{checkpoint-4200 → checkpoint-4600}/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:77f2eb217cfb36ea105abf27909c020d1f994a2f0d0b9529c4f34c95fa1979a6
+oid sha256:7fce4744c6f9f99ebf88811d26dadf3b80e0f50d32d46161fc773f1754ac501b
 size 1001752701

{checkpoint-4200 → checkpoint-4600}/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e39955e63618e46f3bdc48794eb468c4a2a362a14b852408614167c948270a9f
+oid sha256:87cf73a108581136c28003995848a3469b34884b54056edbd96d9426b19286d4
 size 27772

{checkpoint-4200 → checkpoint-4600}/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ace003ac134e276924eb3ea4f065dababf4315826919aa87c57bd6ed1f3c9f8f
+oid sha256:4ed08fdd87a9eb079576d2d2e79f82d8d181e81e5ea1aaaea16afb65f918ed37
 size 27772

{checkpoint-4200 → checkpoint-4600}/rng_state_10.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a7214f06b466b42d304a3bac986a1c2ef8dc14dafe3ca963c50cc13faf825db9
+oid sha256:5231e38f102d1b091797b4cb63ddc5adec05d33a2851436dd68077cbb0064d28
 size 27789

{checkpoint-4200 → checkpoint-4600}/rng_state_11.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9321a3c8b3f28ea39e71d0c9199f896d1989599a3945fb573f479f28b13c7081
+oid sha256:69f01b3fcb68dbb72dd1357f58bcfac99842616b87f82438c24f947602cd918b
 size 27789

{checkpoint-4200 → checkpoint-4600}/rng_state_12.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0adc7008fcd6248e4886234c2141a70e79e8b53bf5fc19aae0deb1b8a5ffbdbd
+oid sha256:3f2bd26cbc1398ca2ebd540399d0c4d6338ed464d548af7c1038f852ca6219db
 size 27789

{checkpoint-4200 → checkpoint-4600}/rng_state_13.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8e9ca26dbc3ab046b9d5dedf18e3e181aa09c97d2ad441ced93c7a1d82f73fca
+oid sha256:858e20e80f50042ced11348b07f992a32041d7732ed58a017be4527ed0a13118
 size 27789

{checkpoint-4200 → checkpoint-4600}/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0bbe635548edc83c591c14a2581c100e77475547bba010c7efba7f187a428d46
+oid sha256:d968dcc8ca9c991c99a196e8d92215f20f736625130c876c9b64041bce288b6d
 size 27772

{checkpoint-4200 → checkpoint-4600}/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:47f9eaccfe6fd830f82205d0a5f025c6394d5f803d79213cd73210a58146d9b0
+oid sha256:9350788e0e90e8024f2d9ed4e04b572e4fb0c1a913ae77e1c4e4f233867609d0
 size 27772

{checkpoint-4200 → checkpoint-4600}/rng_state_4.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:340adbcca40d74a681243ee67ff7bc6c95c55289c128f515419588af41856bcf
+oid sha256:e65ae4aaa3c776537f2bf5c5785311caf3d273bfe047df490befdca5b8e26247
 size 27772

{checkpoint-4200 → checkpoint-4600}/rng_state_5.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fc872f2e697b2a05a8fdeaaef5d8a5579254541515d2738e64317b15f0aea5be
+oid sha256:22bdddfd4edfea8b430f652448271c5f5d87f3505a772758c6417e8ff98655c4
 size 27772

{checkpoint-4200 → checkpoint-4600}/rng_state_6.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d80328e735a3525dddccf939e91d6eca35158a45e34301c42a8db036b696176b
+oid sha256:7a2c49429dcbfeca1faa7a663803a0fdae88ad305b95d23e46226445395b9946
 size 27772

{checkpoint-4200 → checkpoint-4600}/rng_state_7.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4f1edc9d338e7a89fd54ffa70b9782ed3c0abe6fbf326b9254957a771c5923e8
+oid sha256:a9c7570bf926e320eb7895bf959a53e1980cb414ed8bc39b4eda9b9478584053
 size 27772

{checkpoint-4200 → checkpoint-4600}/rng_state_8.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:514fbb84e63510c3d0922a39c6f9a4922b21b3dde59c727d229088232a708135
+oid sha256:fb6fa0fe38db2df9eb256d09e9da98351346be62c7aab5a72c2a011d6913561a
 size 27772

{checkpoint-4200 → checkpoint-4600}/rng_state_9.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c1ee58ff754f86aeb8212b1e7ab704ad3b725ce3348aad288674dee695e18524
+oid sha256:8ed3fe525bbbbe135513154ac6a01e1c556fe2d6d16f9ab7210f0476687c9b5e
 size 27772

{checkpoint-4200 → checkpoint-4600}/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:03d7576e67497d9cef602df64ab49be9ebeda7e0cb4f0d510989a98dd6dae490
+oid sha256:fdc6ea8a6b4cf0316e8365ff2c59039288c5d56c6df930ea3adced9242e298fc
 size 627

{checkpoint-4200 → checkpoint-4600}/trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.6285381930981,
-  "global_step": 4200,
+  "epoch": 1.7836370686312524,
+  "global_step": 4600,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1022,11 +1022,107 @@
       "learning_rate": 8.676433077322215e-05,
       "loss": 0.7956,
       "step": 4200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.575754075832973e-05,
+      "loss": 0.7968,
+      "step": 4225
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.475222217200801e-05,
+      "loss": 0.7905,
+      "step": 4250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.374847887655112e-05,
+      "loss": 0.7889,
+      "step": 4275
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.274641457150543e-05,
+      "loss": 0.7988,
+      "step": 4300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.174613278295608e-05,
+      "loss": 0.7947,
+      "step": 4325
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.074773685283137e-05,
+      "loss": 0.7929,
+      "step": 4350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.97513299282264e-05,
+      "loss": 0.7949,
+      "step": 4375
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.875701495074638e-05,
+      "loss": 0.7925,
+      "step": 4400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.776489464587158e-05,
+      "loss": 0.7917,
+      "step": 4425
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.677507151234448e-05,
+      "loss": 0.7905,
+      "step": 4450
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.578764781158034e-05,
+      "loss": 0.7912,
+      "step": 4475
+    },
+    {
+      "epoch": 1.74,
+      "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
     }
   ],
   "max_steps": 7737,
   "num_train_epochs": 3,
-  "total_flos": 1.8086435753964863e+19,
+  "total_flos": 1.980912103157924e+19,
   "trial_name": null,
   "trial_params": null
 }

{checkpoint-4200 → checkpoint-4600}/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c24a630a19529cd8950cf57e21bd075e18eda756219dfbf6103ddefc1c70630c
+oid sha256:4458bf553277b063d908ed9668b95abc04892c52d8d793f6b007433394d06f17
 size 4027