Training in progress, step 5300

adapter_config.json

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

adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:28c2d0c51dd5ec8f66ff0bbdb8ae6185bd1eca75e6b3b3409367c09446351f5a
+oid sha256:14ca48bd88a31002dc05f9274bc978bc8b5b8ae3e51470e74ee7bc0bb12284e9
 size 500897101

{checkpoint-4900 → checkpoint-5200}/adapter_model/adapter_config.json

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

{checkpoint-4900 → checkpoint-5200/adapter_model}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2a3bb33a0c4878b032aec2026e3879865fa62989d9939826a716eb3580fa34cd
+oid sha256:28c2d0c51dd5ec8f66ff0bbdb8ae6185bd1eca75e6b3b3409367c09446351f5a
 size 500897101

{checkpoint-4900 → checkpoint-5300}/adapter_config.json

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

{checkpoint-4900/adapter_model → checkpoint-5300}/adapter_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2a3bb33a0c4878b032aec2026e3879865fa62989d9939826a716eb3580fa34cd
+oid sha256:14ca48bd88a31002dc05f9274bc978bc8b5b8ae3e51470e74ee7bc0bb12284e9
 size 500897101

{checkpoint-4900 → checkpoint-5300}/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:89baee7be101e0977ca1b6a39db7917193704b00b2bc0dc4bbf1b2e60e7e6a22
+oid sha256:869a51ca719bee8b6aab465be97fa4ac5e228c769321c2712f644067ceeca076
 size 1001752701

{checkpoint-4900 → checkpoint-5300}/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3d2fb5eb561d4b92c89062c2939861ac27d199067133d2975e1e2aecd6254425
+oid sha256:9267467db4bfc0b62f4b2992b98c478568d6b740025cdb5016f4102da1504163
 size 27772

{checkpoint-4900 → checkpoint-5300}/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:845f7f7308d4eb8331386ad93aebcacf5679b5a7f2b26d79e0e881c33768d801
+oid sha256:ea0db0e586ec1c1e547243b2e02d8b0d760f3d113e51420b567322fb80b86283
 size 27772

{checkpoint-4900 → checkpoint-5300}/rng_state_10.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ac19165b3d0d8492404b2597a692a25174dc25bbebee0818cc70fb91e7fa66a9
+oid sha256:fcfb12512c0be896f0842434c14afaa23ede7d6aadbd83bfd18570859fcdecdd
 size 27789

{checkpoint-4900 → checkpoint-5300}/rng_state_11.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8ba175f65490c813c43be1b8e90c05273b365a5fc35034a39d3e40171b6649d8
+oid sha256:fcd57397386863534e25c2052e0eb5a3986965528e8332bead4e6da6c8a52a60
 size 27789

{checkpoint-4900 → checkpoint-5300}/rng_state_12.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ff1e6d6959816301feff7a9806d0ca028fe2a211e238b9dca8fd390289758baa
+oid sha256:3b494220b086ab33c457cd9cb3298d5707ae27861e1a1ea15cf7b2846c34edb3
 size 27789

{checkpoint-4900 → checkpoint-5300}/rng_state_13.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f1211e56e1d3e784cd9f700dcd2e4b8d94627474333f9ea95573cbff916f6590
+oid sha256:66ddbbc44b73707a46dbe70a821775b9215bfb3188b97867dfd5d788f2c4bd1b
 size 27789

{checkpoint-4900 → checkpoint-5300}/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1a4a16265fbd33a987004f285a9ac8e9d11fa74e2f1e1ebfae43b07d355feb1a
+oid sha256:e09271dc29724c758580e8e6864e76555b7a1dd95ca5d0036f08cdd9f67fbaec
 size 27772

{checkpoint-4900 → checkpoint-5300}/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:25be8bdc125a49f8faa5491285fe8c51cbd061cd60a12c31e1d294969eaeed1b
+oid sha256:65b0c519b8283902066db24d39bebbf0ec0f57d632779c56be9c9d58f7cdb78e
 size 27772

{checkpoint-4900 → checkpoint-5300}/rng_state_4.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f8edc43261484ceb88153a0cf622a2f1d79ae2bd2a22d2cfc258fbc4d67c9fcf
+oid sha256:868a6a26d82f017652ee4b9e21c96ec36ab5f96be4eeedf65e27eda71ff93f25
 size 27772

{checkpoint-4900 → checkpoint-5300}/rng_state_5.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f3c01adf24a8fb5ced49e768a00799034bf23fdbe170faf373d7bffc71cf330e
+oid sha256:d82c47d40cdc9cbed8bc1b1f8fdcb354fb8b6ac525cfce07fab2d1af5791c195
 size 27772

{checkpoint-4900 → checkpoint-5300}/rng_state_6.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1431fdce90c9dffa41af7e45225d7e937f8a892f36234dd2c9c0c5ea825d344b
+oid sha256:64d8ec328ebab19d4fa69241bbb5de66af7b09aa5f0d71d9e37be38a5170d24e
 size 27772

{checkpoint-4900 → checkpoint-5300}/rng_state_7.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:698169f9adbf09c9e392876211e57228599d852964332781ccccb91b02c99668
+oid sha256:cf9d89caef032992b8f5efcdaf7ab987e37e4fb52ec8d47520f814092e1e7ab3
 size 27772

{checkpoint-4900 → checkpoint-5300}/rng_state_8.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1e12b62affc946cb152ae7675f1074ed4a8c28a930020230d21a02278a8cfc66
+oid sha256:8913ebfea32419c486eaa9895c49a7962a44933cd37f4965710a3560fe737c25
 size 27772

{checkpoint-4900 → checkpoint-5300}/rng_state_9.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8d0857929e84e3ee4f8ead280fd3f99826ceb6e6e749cce250845a12d23d0cc7
+oid sha256:685f43d3417ce77333bd61cb9e79c3d6e0421bb659bae8689caafc6c6b0d8752
 size 27772

{checkpoint-4900 → checkpoint-5300}/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fce2b4a13969f53fdf68394680e1b829aa39975e2753c73d5d093518e2315b1c
+oid sha256:0d904fb502ced275ef5e7925781f477f4c89d48daa60e9f45b9339adf1aaa77d
 size 627

{checkpoint-4900 → checkpoint-5300}/trainer_state.json

@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.8999612252811167,
-  "global_step": 4900,
+  "epoch": 2.055060100814269,
+  "global_step": 5300,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1190,11 +1190,115 @@
       "learning_rate": 5.946628610250484e-05,
       "loss": 0.7918,
       "step": 4900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.853920568929996e-05,
+      "loss": 0.7921,
+      "step": 4925
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.761640870750799e-05,
+      "loss": 0.7878,
+      "step": 4950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.669799049388375e-05,
+      "loss": 0.7901,
+      "step": 4975
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.578404593279911e-05,
+      "loss": 0.7858,
+      "step": 5000
+    },
+    {
+      "epoch": 1.94,
+      "eval_loss": 0.807844877243042,
+      "eval_runtime": 59.586,
+      "eval_samples_per_second": 12.251,
+      "eval_steps_per_second": 0.889,
+      "step": 5000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.487466944644033e-05,
+      "loss": 0.7902,
+      "step": 5025
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.3969954985052996e-05,
+      "loss": 0.7979,
+      "step": 5050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.306999601723579e-05,
+      "loss": 0.7931,
+      "step": 5075
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.21748855202839e-05,
+      "loss": 0.7868,
+      "step": 5100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.128471597058342e-05,
+      "loss": 0.7993,
+      "step": 5125
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.03995793340572e-05,
+      "loss": 0.7892,
+      "step": 5150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9519567056663694e-05,
+      "loss": 0.7788,
+      "step": 5175
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.864477005494938e-05,
+      "loss": 0.7654,
+      "step": 5200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.777527870665592e-05,
+      "loss": 0.7468,
+      "step": 5225
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.691118284138296e-05,
+      "loss": 0.7359,
+      "step": 5250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.605257173130763e-05,
+      "loss": 0.7422,
+      "step": 5275
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.519953408196152e-05,
+      "loss": 0.7424,
+      "step": 5300
     }
   ],
   "max_steps": 7737,
   "num_train_epochs": 3,
-  "total_flos": 2.1101658419601867e+19,
+  "total_flos": 2.282482987033428e+19,
   "trial_name": null,
   "trial_params": null
 }

{checkpoint-4900 → checkpoint-5300}/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:354c4769acaf21317d8ab58ba2dd133763f3bee1aec7488fb334e45f3e80ebc3
+oid sha256:689d52379bcc7c50e04c40b22a97b473b8de3f17b4096bebf81eb9f37e1dafa6
 size 4027

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b5cc18faa1420e425c2fed06bfe2dd967461487c15531bd94429b7a3c0c02a49
+oid sha256:689d52379bcc7c50e04c40b22a97b473b8de3f17b4096bebf81eb9f37e1dafa6
 size 4027