2040

.gitattributes

@@ -19,3 +19,5 @@ global_step1700/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt filter=lfs diff=l
 global_step1700/mp_rank_00_model_states.pt filter=lfs diff=lfs merge=lfs -text
 global_step1870/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt filter=lfs diff=lfs merge=lfs -text
 global_step1870/mp_rank_00_model_states.pt filter=lfs diff=lfs merge=lfs -text
+global_step2040/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt filter=lfs diff=lfs merge=lfs -text
+global_step2040/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 sha256:93145fb67a6755f3358da8d36be3e02948c85fb7102b8b9e3b481f8674852c53
+oid sha256:16971df3fe099a7bd790d6216dabc54390279e71f1621e23083a0ab16941a804
 size 335606536

{global_step1870 → global_step2040}/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:49c974b0eae7df4d70fa3958f5269a525f7333e726267a4b9a59af45154dccc2
+oid sha256:2a4846acc76d4b43796d990ec2143586786a859c5c16dc1c658341837d39b08e
 size 2013320965

{global_step1870 → global_step2040}/mp_rank_00_model_states.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9844d738eb2fa46e0ae4955e3c82502b358271721cc524822f0ac9da6d4c9639
+oid sha256:ba56bb99b2201507af3c1e408629fe38100bfd6155c53d5bfd8fe299eae49a28
 size 335901529

latest

@@ -1 +1 @@
-global_step1870
+global_step2040

trainer_state.json

@@ -1,10 +1,10 @@
 {
-  "best_global_step": 1870,
-  "best_metric": 0.31162187457084656,
-  "best_model_checkpoint": "/home/longtail/data/outputs/exaone_7b_lora_rtlcoder_ratio1.0/checkpoint-1870",
-  "epoch": 1.409270774448841,
+  "best_global_step": 2040,
+  "best_metric": 0.3090449571609497,
+  "best_model_checkpoint": "/home/longtail/data/outputs/exaone_7b_lora_rtlcoder_ratio1.0/checkpoint-2040",
+  "epoch": 1.5374034294328247,
   "eval_steps": 170,
-  "global_step": 1870,
+  "global_step": 2040,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1999,6 +1999,187 @@
       "eval_samples_per_second": 5.969,
       "eval_steps_per_second": 1.494,
       "step": 1870
+    },
+    {
+      "entropy": 0.2755126953125,
+      "epoch": 1.416807989447899,
+      "grad_norm": 0.34847527742385864,
+      "learning_rate": 1.6824445452838112e-05,
+      "loss": 0.2797748327255249,
+      "mean_token_accuracy": 0.9122792810201645,
+      "num_tokens": 23751843.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3048583984375,
+      "epoch": 1.424345204446957,
+      "grad_norm": 0.34166133403778076,
+      "learning_rate": 1.6788679612845275e-05,
+      "loss": 0.3143250226974487,
+      "mean_token_accuracy": 0.9053974837064743,
+      "num_tokens": 23890129.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.2959228515625,
+      "epoch": 1.4318824194460147,
+      "grad_norm": 0.4138602316379547,
+      "learning_rate": 1.6752751969657364e-05,
+      "loss": 0.3098402738571167,
+      "mean_token_accuracy": 0.9060688123106957,
+      "num_tokens": 24025639.0,
+      "step": 1900
+    },
+    {
+      "entropy": 0.2892822265625,
+      "epoch": 1.4394196344450725,
+      "grad_norm": 0.35475072264671326,
+      "learning_rate": 1.6716663379583373e-05,
+      "loss": 0.30137479305267334,
+      "mean_token_accuracy": 0.9065017476677895,
+      "num_tokens": 24142755.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.2736083984375,
+      "epoch": 1.4469568494441303,
+      "grad_norm": 0.38053274154663086,
+      "learning_rate": 1.6680414702768358e-05,
+      "loss": 0.2755943775177002,
+      "mean_token_accuracy": 0.9128162145614624,
+      "num_tokens": 24271034.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.2882568359375,
+      "epoch": 1.4544940644431883,
+      "grad_norm": 0.3685821294784546,
+      "learning_rate": 1.6644006803172926e-05,
+      "loss": 0.29870429039001467,
+      "mean_token_accuracy": 0.9083519443869591,
+      "num_tokens": 24401299.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.2899169921875,
+      "epoch": 1.462031279442246,
+      "grad_norm": 0.38053128123283386,
+      "learning_rate": 1.660744054855263e-05,
+      "loss": 0.2999034643173218,
+      "mean_token_accuracy": 0.9092144444584846,
+      "num_tokens": 24525590.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.3008544921875,
+      "epoch": 1.4695684944413039,
+      "grad_norm": 0.3747199773788452,
+      "learning_rate": 1.657071681043731e-05,
+      "loss": 0.3081289052963257,
+      "mean_token_accuracy": 0.9040897369384766,
+      "num_tokens": 24652105.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.28818359375,
+      "epoch": 1.477105709440362,
+      "grad_norm": 0.4361964762210846,
+      "learning_rate": 1.6533836464110303e-05,
+      "loss": 0.2913468599319458,
+      "mean_token_accuracy": 0.9085044726729393,
+      "num_tokens": 24781538.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.2942626953125,
+      "epoch": 1.4846429244394197,
+      "grad_norm": 0.4351835548877716,
+      "learning_rate": 1.649680038858759e-05,
+      "loss": 0.30144288539886477,
+      "mean_token_accuracy": 0.9078996136784554,
+      "num_tokens": 24900040.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.309912109375,
+      "epoch": 1.4921801394384775,
+      "grad_norm": 0.3614610433578491,
+      "learning_rate": 1.645960946659685e-05,
+      "loss": 0.32563717365264894,
+      "mean_token_accuracy": 0.9042312502861023,
+      "num_tokens": 25029356.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.288232421875,
+      "epoch": 1.4997173544375353,
+      "grad_norm": 0.46183428168296814,
+      "learning_rate": 1.64222645845564e-05,
+      "loss": 0.2938676118850708,
+      "mean_token_accuracy": 0.9088562294840813,
+      "num_tokens": 25150702.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.287841796875,
+      "epoch": 1.507254569436593,
+      "grad_norm": 0.35748639702796936,
+      "learning_rate": 1.638476663255409e-05,
+      "loss": 0.3016012907028198,
+      "mean_token_accuracy": 0.90945535749197,
+      "num_tokens": 25289821.0,
+      "step": 2000
+    },
+    {
+      "entropy": 0.29658203125,
+      "epoch": 1.5147917844356509,
+      "grad_norm": 0.392511785030365,
+      "learning_rate": 1.6347116504326082e-05,
+      "loss": 0.30182077884674074,
+      "mean_token_accuracy": 0.9069562748074531,
+      "num_tokens": 25427874.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.2969482421875,
+      "epoch": 1.522328999434709,
+      "grad_norm": 0.4357432425022125,
+      "learning_rate": 1.630931509723554e-05,
+      "loss": 0.30153517723083495,
+      "mean_token_accuracy": 0.9072390154004097,
+      "num_tokens": 25554351.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.2900390625,
+      "epoch": 1.5298662144337667,
+      "grad_norm": 0.36660265922546387,
+      "learning_rate": 1.6271363312251253e-05,
+      "loss": 0.30401484966278075,
+      "mean_token_accuracy": 0.909621711075306,
+      "num_tokens": 25687339.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.291259765625,
+      "epoch": 1.5374034294328247,
+      "grad_norm": 0.3444574177265167,
+      "learning_rate": 1.6233262053926165e-05,
+      "loss": 0.29833531379699707,
+      "mean_token_accuracy": 0.9089738965034485,
+      "num_tokens": 25819705.0,
+      "step": 2040
+    },
+    {
+      "epoch": 1.5374034294328247,
+      "eval_entropy": 0.29168186417545183,
+      "eval_loss": 0.3090449571609497,
+      "eval_mean_token_accuracy": 0.9057694819856839,
+      "eval_num_tokens": 25819705.0,
+      "eval_runtime": 444.9473,
+      "eval_samples_per_second": 5.963,
+      "eval_steps_per_second": 1.492,
+      "step": 2040
     }
   ],
   "logging_steps": 10,
@@ -2018,7 +2199,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.787706633963438e+18,
+  "total_flos": 1.9547214430527816e+18,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null