Training in progress, step 25, checkpoint

last-checkpoint/adapter_config.json

@@ -31,10 +31,10 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "qkv_proj",
-    "o_proj",
+    "down_proj",
     "gate_up_proj",
-    "down_proj"
+    "o_proj",
+    "qkv_proj"
   ],
   "target_parameters": null,
   "task_type": "CAUSAL_LM",

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:07ca27b199f09823d364b1800a536a7081900016c2bcc303ce1bac8c2764301c
+oid sha256:57e4734cbbe72f9e9bae98c394fc3d207a4fd1048611a77cfce5abf5553b06db
 size 50366280

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:550be29018b76f393595d97771205281a5d4d2977ef5cdf1b87b832630484a2b
+oid sha256:d22baff8ed7ca9cbeae04980dd00f21d06b23696536eb466e87398ce92d167ee
 size 51423589

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5df20225b78134a4de357342e5cf67eae7b29d2a1c35147895dfbfa5d5b84eaa
+oid sha256:d81e990436bcd40e045db972dfd5d57a4c9c7a635f60b749fe3d5de220b821dd
 size 14645

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fe0236d6a2fed952f5fc134eee26e92c7aa46bdf4f7ec73861bc370d83d8a762
+oid sha256:fd65659e9d4065cb7ced7e5dc6a84ce910bd21d03812af593d819f8f600f4322
 size 1465

last-checkpoint/trainer_state.json

@@ -1,31 +1,281 @@
 {
-  "best_global_step": 5,
-  "best_metric": 2.391779661178589,
-  "best_model_checkpoint": "./outputs/phi3-clause-lora-dryrun/checkpoint-5",
-  "epoch": 2.8,
-  "eval_steps": 100,
-  "global_step": 5,
+  "best_global_step": 25,
+  "best_metric": 1.4348562955856323,
+  "best_model_checkpoint": "./outputs/phi3-clause-lora-pushcheck/checkpoint-25",
+  "epoch": 0.13360053440213762,
+  "eval_steps": 25,
+  "global_step": 25,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 2.8,
-      "eval_entropy": 1.8215465545654297,
-      "eval_loss": 2.391779661178589,
-      "eval_mean_token_accuracy": 0.5219630599021912,
-      "eval_num_tokens": 9298.0,
-      "eval_runtime": 1.1713,
-      "eval_samples_per_second": 8.538,
-      "eval_steps_per_second": 1.708,
+      "entropy": 1.3057959824800491,
+      "epoch": 0.005344021376085505,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.0,
+      "loss": 2.3924810886383057,
+      "mean_token_accuracy": 0.5911420322954655,
+      "num_tokens": 2200.0,
+      "step": 1
+    },
+    {
+      "entropy": 1.5550646409392357,
+      "epoch": 0.01068804275217101,
+      "grad_norm": 2.84375,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 2.9143970012664795,
+      "mean_token_accuracy": 0.5189028866589069,
+      "num_tokens": 4005.0,
+      "step": 2
+    },
+    {
+      "entropy": 1.4377547204494476,
+      "epoch": 0.01603206412825651,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 2.368616819381714,
+      "mean_token_accuracy": 0.5810851827263832,
+      "num_tokens": 6480.0,
+      "step": 3
+    },
+    {
+      "entropy": 1.2220652177929878,
+      "epoch": 0.02137608550434202,
+      "grad_norm": 1.5703125,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 1.7445847988128662,
+      "mean_token_accuracy": 0.7135058641433716,
+      "num_tokens": 8441.0,
+      "step": 4
+    },
+    {
+      "entropy": 1.3147591650485992,
+      "epoch": 0.026720106880427523,
+      "grad_norm": 1.828125,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 2.425869941711426,
+      "mean_token_accuracy": 0.5845252685248852,
+      "num_tokens": 10437.0,
       "step": 5
+    },
+    {
+      "entropy": 1.3565818294882774,
+      "epoch": 0.03206412825651302,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 1.7197703123092651,
+      "mean_token_accuracy": 0.6920341700315475,
+      "num_tokens": 12134.0,
+      "step": 6
+    },
+    {
+      "entropy": 1.2649112790822983,
+      "epoch": 0.03740814963259853,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00010909090909090909,
+      "loss": 1.4319822788238525,
+      "mean_token_accuracy": 0.6885706260800362,
+      "num_tokens": 14525.0,
+      "step": 7
+    },
+    {
+      "entropy": 1.60972660779953,
+      "epoch": 0.04275217100868404,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00012727272727272728,
+      "loss": 1.8486095666885376,
+      "mean_token_accuracy": 0.6440986394882202,
+      "num_tokens": 16336.0,
+      "step": 8
+    },
+    {
+      "entropy": 1.4543273150920868,
+      "epoch": 0.04809619238476954,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00014545454545454546,
+      "loss": 1.575211524963379,
+      "mean_token_accuracy": 0.7021859884262085,
+      "num_tokens": 18279.0,
+      "step": 9
+    },
+    {
+      "entropy": 1.567221999168396,
+      "epoch": 0.053440213760855046,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00016363636363636366,
+      "loss": 1.4245097637176514,
+      "mean_token_accuracy": 0.6823129132390022,
+      "num_tokens": 20514.0,
+      "step": 10
+    },
+    {
+      "entropy": 1.5797198563814163,
+      "epoch": 0.058784235136940546,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 1.264983892440796,
+      "mean_token_accuracy": 0.7505967691540718,
+      "num_tokens": 22207.0,
+      "step": 11
+    },
+    {
+      "entropy": 1.8289320468902588,
+      "epoch": 0.06412825651302605,
+      "grad_norm": 2.125,
+      "learning_rate": 0.0002,
+      "loss": 1.5185409784317017,
+      "mean_token_accuracy": 0.6885010972619057,
+      "num_tokens": 24341.0,
+      "step": 12
+    },
+    {
+      "entropy": 1.7916453033685684,
+      "epoch": 0.06947227788911156,
+      "grad_norm": 1.7734375,
+      "learning_rate": 0.00019863613034027224,
+      "loss": 1.3594837188720703,
+      "mean_token_accuracy": 0.7468798607587814,
+      "num_tokens": 26821.0,
+      "step": 13
+    },
+    {
+      "entropy": 1.350630447268486,
+      "epoch": 0.07481629926519706,
+      "grad_norm": 2.15625,
+      "learning_rate": 0.00019458172417006347,
+      "loss": 1.0361241102218628,
+      "mean_token_accuracy": 0.7759386524558067,
+      "num_tokens": 28605.0,
+      "step": 14
+    },
+    {
+      "entropy": 1.53353850543499,
+      "epoch": 0.08016032064128256,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.0001879473751206489,
+      "loss": 1.2417271137237549,
+      "mean_token_accuracy": 0.7104115560650826,
+      "num_tokens": 30698.0,
+      "step": 15
+    },
+    {
+      "entropy": 1.0438459292054176,
+      "epoch": 0.08550434201736808,
+      "grad_norm": 1.78125,
+      "learning_rate": 0.00017891405093963938,
+      "loss": 0.7792830467224121,
+      "mean_token_accuracy": 0.8171160146594048,
+      "num_tokens": 32595.0,
+      "step": 16
+    },
+    {
+      "entropy": 1.1498453691601753,
+      "epoch": 0.09084836339345358,
+      "grad_norm": 2.171875,
+      "learning_rate": 0.00016772815716257412,
+      "loss": 0.8656947016716003,
+      "mean_token_accuracy": 0.7883986905217171,
+      "num_tokens": 34179.0,
+      "step": 17
+    },
+    {
+      "entropy": 0.9672329127788544,
+      "epoch": 0.09619238476953908,
+      "grad_norm": 1.96875,
+      "learning_rate": 0.00015469481581224272,
+      "loss": 0.8689895868301392,
+      "mean_token_accuracy": 0.8031240478157997,
+      "num_tokens": 36332.0,
+      "step": 18
+    },
+    {
+      "entropy": 0.7483968064188957,
+      "epoch": 0.10153640614562458,
+      "grad_norm": 2.140625,
+      "learning_rate": 0.00014016954246529696,
+      "loss": 0.5695925354957581,
+      "mean_token_accuracy": 0.8222300335764885,
+      "num_tokens": 38903.0,
+      "step": 19
+    },
+    {
+      "entropy": 0.7375008687376976,
+      "epoch": 0.10688042752171009,
+      "grad_norm": 2.78125,
+      "learning_rate": 0.00012454854871407994,
+      "loss": 0.6309845447540283,
+      "mean_token_accuracy": 0.8498999550938606,
+      "num_tokens": 41314.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6442421264946461,
+      "epoch": 0.11222444889779559,
+      "grad_norm": 2.375,
+      "learning_rate": 0.00010825793454723325,
+      "loss": 0.6821945905685425,
+      "mean_token_accuracy": 0.8697243854403496,
+      "num_tokens": 43552.0,
+      "step": 21
+    },
+    {
+      "entropy": 0.6835456639528275,
+      "epoch": 0.11756847027388109,
+      "grad_norm": 2.328125,
+      "learning_rate": 9.174206545276677e-05,
+      "loss": 0.5624313950538635,
+      "mean_token_accuracy": 0.8455757722258568,
+      "num_tokens": 45397.0,
+      "step": 22
+    },
+    {
+      "entropy": 0.689422957599163,
+      "epoch": 0.1229124916499666,
+      "grad_norm": 2.546875,
+      "learning_rate": 7.54514512859201e-05,
+      "loss": 0.48625773191452026,
+      "mean_token_accuracy": 0.8459644466638565,
+      "num_tokens": 47642.0,
+      "step": 23
+    },
+    {
+      "entropy": 0.586566150188446,
+      "epoch": 0.1282565130260521,
+      "grad_norm": 2.609375,
+      "learning_rate": 5.983045753470308e-05,
+      "loss": 0.45920994877815247,
+      "mean_token_accuracy": 0.8874738067388535,
+      "num_tokens": 49571.0,
+      "step": 24
+    },
+    {
+      "entropy": 0.5189335495233536,
+      "epoch": 0.13360053440213762,
+      "grad_norm": 3.40625,
+      "learning_rate": 4.530518418775733e-05,
+      "loss": 0.6368916630744934,
+      "mean_token_accuracy": 0.8659464195370674,
+      "num_tokens": 52032.0,
+      "step": 25
+    },
+    {
+      "epoch": 0.13360053440213762,
+      "eval_entropy": 0.784009460591327,
+      "eval_loss": 1.4348562955856323,
+      "eval_mean_token_accuracy": 0.783792299234261,
+      "eval_num_tokens": 52032.0,
+      "eval_runtime": 428.8917,
+      "eval_samples_per_second": 6.603,
+      "eval_steps_per_second": 0.825,
+      "step": 25
     }
   ],
-  "logging_steps": 25,
-  "max_steps": 5,
+  "logging_steps": 1,
+  "max_steps": 30,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 3,
-  "save_steps": 100,
+  "num_train_epochs": 1,
+  "save_steps": 25,
   "stateful_callbacks": {
     "TrainerControl": {
       "args": {
@@ -33,12 +283,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": true
+        "should_training_stop": false
       },
       "attributes": {}
     }
   },
-  "total_flos": 258099686682624.0,
+  "total_flos": 1509696529403904.0,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ea74cab36abf68a311e6119ef69bfc41d792ba29a0e224d1d7c176cf5d210fc1
+oid sha256:bde0e973d79d797093a52f50066fc862dfb658ee6bc91d84d4fa39f3d3f84dc6
 size 5777