850

.gitattributes

@@ -5,3 +5,5 @@ global_step510/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt filter=lfs diff=lf
 global_step510/mp_rank_00_model_states.pt filter=lfs diff=lfs merge=lfs -text
 global_step680/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt filter=lfs diff=lfs merge=lfs -text
 global_step680/mp_rank_00_model_states.pt filter=lfs diff=lfs merge=lfs -text
+global_step850/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt filter=lfs diff=lfs merge=lfs -text
+global_step850/mp_rank_00_model_states.pt filter=lfs diff=lfs merge=lfs -text

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2ea15be39be1837fa21589b77816d9254349084c30293939fa0f4709cc3485fd
+oid sha256:c565ad20c17047fcc44c03358debdd81a1912ea5db26cc162611de817111acc1
 size 335606536

{global_step680 → global_step850}/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8178f9d052b7f7e11805badbebeeea03c2ffa515ea76367e29ce096f27cf13cc
+oid sha256:593d8c1cc39f7025dccdbaa7881821a383e3313a9e8f02cd2a43e4725acb3a7c
 size 2013320965

{global_step680 → global_step850}/mp_rank_00_model_states.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:08c547515933f9191a08bbcdde14b6b1347d89163eab34bc2a2426d8e83f3fb5
+oid sha256:e1653ada7c6fcb7465a62e90ddb705e2c283ef5c864af8000b3ed9bd1d4f71a7
 size 335901529

latest

@@ -1 +1 @@
-global_step680
+global_step850

trainer_state.json

@@ -1,10 +1,10 @@
 {
-  "best_global_step": 680,
-  "best_metric": 0.3412366807460785,
-  "best_model_checkpoint": "/home/longtail/data/outputs/exaone_7b_lora_rtlcoder_ratio1.0/checkpoint-680",
-  "epoch": 0.5125306199359336,
+  "best_global_step": 850,
+  "best_metric": 0.33324772119522095,
+  "best_model_checkpoint": "/home/longtail/data/outputs/exaone_7b_lora_rtlcoder_ratio1.0/checkpoint-850",
+  "epoch": 0.640663274919917,
   "eval_steps": 170,
-  "global_step": 680,
+  "global_step": 850,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -732,6 +732,187 @@
       "eval_samples_per_second": 5.959,
       "eval_steps_per_second": 1.491,
       "step": 680
+    },
+    {
+      "entropy": 0.3375244140625,
+      "epoch": 0.5200678349349915,
+      "grad_norm": 0.39377152919769287,
+      "learning_rate": 1.971638660083016e-05,
+      "loss": 0.3446618318557739,
+      "mean_token_accuracy": 0.8973624289035798,
+      "num_tokens": 8734967.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.380859375,
+      "epoch": 0.5276050499340493,
+      "grad_norm": 0.3425155282020569,
+      "learning_rate": 1.970472626939742e-05,
+      "loss": 0.3928657293319702,
+      "mean_token_accuracy": 0.8857068896293641,
+      "num_tokens": 8866789.0,
+      "step": 700
+    },
+    {
+      "entropy": 0.3181884765625,
+      "epoch": 0.5351422649331072,
+      "grad_norm": 0.39210009574890137,
+      "learning_rate": 1.969283463293818e-05,
+      "loss": 0.3279300928115845,
+      "mean_token_accuracy": 0.901725186407566,
+      "num_tokens": 8989637.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.3260986328125,
+      "epoch": 0.542679479932165,
+      "grad_norm": 0.36499711871147156,
+      "learning_rate": 1.9680711974880868e-05,
+      "loss": 0.3395835399627686,
+      "mean_token_accuracy": 0.8985119208693504,
+      "num_tokens": 9107079.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.3133056640625,
+      "epoch": 0.5502166949312229,
+      "grad_norm": 0.3930191099643707,
+      "learning_rate": 1.9668358584160136e-05,
+      "loss": 0.3265484094619751,
+      "mean_token_accuracy": 0.9016185060143471,
+      "num_tokens": 9227783.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.340673828125,
+      "epoch": 0.5577539099302807,
+      "grad_norm": 0.3877074122428894,
+      "learning_rate": 1.965577475520999e-05,
+      "loss": 0.34572837352752683,
+      "mean_token_accuracy": 0.8980929121375084,
+      "num_tokens": 9352288.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.3293212890625,
+      "epoch": 0.5652911249293386,
+      "grad_norm": 0.3847011625766754,
+      "learning_rate": 1.964296078795675e-05,
+      "loss": 0.33631391525268556,
+      "mean_token_accuracy": 0.8981146275997162,
+      "num_tokens": 9475875.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.3515380859375,
+      "epoch": 0.5728283399283964,
+      "grad_norm": 0.3962864577770233,
+      "learning_rate": 1.9629916987811924e-05,
+      "loss": 0.3633269309997559,
+      "mean_token_accuracy": 0.8934466958045959,
+      "num_tokens": 9601762.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.31484375,
+      "epoch": 0.5803655549274543,
+      "grad_norm": 0.3556186556816101,
+      "learning_rate": 1.961664366566491e-05,
+      "loss": 0.3226339101791382,
+      "mean_token_accuracy": 0.9032136023044586,
+      "num_tokens": 9729416.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.34990234375,
+      "epoch": 0.5879027699265121,
+      "grad_norm": 0.3913438618183136,
+      "learning_rate": 1.9603141137875596e-05,
+      "loss": 0.358534836769104,
+      "mean_token_accuracy": 0.8945465311408043,
+      "num_tokens": 9862899.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.343603515625,
+      "epoch": 0.59543998492557,
+      "grad_norm": 0.39869415760040283,
+      "learning_rate": 1.9589409726266822e-05,
+      "loss": 0.3582808494567871,
+      "mean_token_accuracy": 0.8956417754292488,
+      "num_tokens": 9990936.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.3222900390625,
+      "epoch": 0.6029771999246278,
+      "grad_norm": 0.46191906929016113,
+      "learning_rate": 1.9575449758116703e-05,
+      "loss": 0.32662172317504884,
+      "mean_token_accuracy": 0.9030146107077599,
+      "num_tokens": 10124909.0,
+      "step": 800
+    },
+    {
+      "entropy": 0.3161376953125,
+      "epoch": 0.6105144149236857,
+      "grad_norm": 0.35114115476608276,
+      "learning_rate": 1.956126156615083e-05,
+      "loss": 0.3295163631439209,
+      "mean_token_accuracy": 0.8997720777988434,
+      "num_tokens": 10252655.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.335595703125,
+      "epoch": 0.6180516299227435,
+      "grad_norm": 0.435343861579895,
+      "learning_rate": 1.9546845488534347e-05,
+      "loss": 0.34328203201293944,
+      "mean_token_accuracy": 0.8988629937171936,
+      "num_tokens": 10378988.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.329248046875,
+      "epoch": 0.6255888449218014,
+      "grad_norm": 0.3690928518772125,
+      "learning_rate": 1.953220186886388e-05,
+      "loss": 0.3510568380355835,
+      "mean_token_accuracy": 0.897222849726677,
+      "num_tokens": 10503801.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.325,
+      "epoch": 0.6331260599208592,
+      "grad_norm": 0.33731427788734436,
+      "learning_rate": 1.9517331056159353e-05,
+      "loss": 0.3301519870758057,
+      "mean_token_accuracy": 0.9016004085540772,
+      "num_tokens": 10632605.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.3230224609375,
+      "epoch": 0.640663274919917,
+      "grad_norm": 0.3374157249927521,
+      "learning_rate": 1.9502233404855672e-05,
+      "loss": 0.335455584526062,
+      "mean_token_accuracy": 0.8995656460523606,
+      "num_tokens": 10758290.0,
+      "step": 850
+    },
+    {
+      "epoch": 0.640663274919917,
+      "eval_entropy": 0.32338926016566266,
+      "eval_loss": 0.33324772119522095,
+      "eval_mean_token_accuracy": 0.8999460333262581,
+      "eval_num_tokens": 10758290.0,
+      "eval_runtime": 445.1709,
+      "eval_samples_per_second": 5.96,
+      "eval_steps_per_second": 1.492,
+      "step": 850
     }
   ],
   "logging_steps": 10,
@@ -751,7 +932,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 6.54336223678038e+17,
+  "total_flos": 8.152533469012951e+17,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null