Model save

README.md

@@ -1,11 +1,9 @@
 ---
 base_model: Qwen/Qwen2.5-1.5B-Instruct
-datasets: HuggingFaceH4/Bespoke-Stratos-17k
 library_name: transformers
 model_name: Qwen2.5-1.5B-Open-R1-Distill
 tags:
 - generated_from_trainer
-- open-r1
 - trl
 - sft
 licence: license
@@ -13,7 +11,7 @@ licence: license
 
 # Model Card for Qwen2.5-1.5B-Open-R1-Distill
 
-This model is a fine-tuned version of [Qwen/Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct) on the [HuggingFaceH4/Bespoke-Stratos-17k](https://huggingface.co/datasets/HuggingFaceH4/Bespoke-Stratos-17k) dataset.
+This model is a fine-tuned version of [Qwen/Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct).
 It has been trained using [TRL](https://github.com/huggingface/trl).
 
 ## Quick start
@@ -29,14 +27,14 @@ print(output["generated_text"])
 
 ## Training procedure
 
-[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/1653401183-mingmingai/huggingface/runs/0i0p6fk9) 
+[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/1653401183-mingmingai/huggingface/runs/batphcqf) 
 
 
 This model was trained with SFT.
 
 ### Framework versions
 
-- TRL: 0.15.0.dev0
+- TRL: 0.16.0.dev0
 - Transformers: 4.49.0.dev0
 - Pytorch: 2.5.1
 - Datasets: 3.2.0

all_results.json

@@ -1,14 +1,8 @@
 {
-    "epoch": 1.0,
-    "eval_loss": 0.7803335785865784,
-    "eval_runtime": 0.8401,
-    "eval_samples": 100,
-    "eval_samples_per_second": 152.364,
-    "eval_steps_per_second": 4.761,
-    "total_flos": 76916824473600.0,
-    "train_loss": 0.8026207946461333,
-    "train_runtime": 498.3301,
+    "total_flos": 76973799899136.0,
+    "train_loss": 0.8025032683942445,
+    "train_runtime": 710.0203,
     "train_samples": 16610,
-    "train_samples_per_second": 43.365,
-    "train_steps_per_second": 0.339
+    "train_samples_per_second": 30.454,
+    "train_steps_per_second": 0.238
 }

config.json

@@ -19,11 +19,11 @@
   "rms_norm_eps": 1e-06,
   "rope_scaling": null,
   "rope_theta": 1000000.0,
-  "sliding_window": null,
+  "sliding_window": 32768,
   "tie_word_embeddings": true,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.49.0.dev0",
-  "use_cache": true,
+  "use_cache": false,
   "use_sliding_window": false,
   "vocab_size": 151936
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:21e70655225165df770f132bba556d7d5e6ebfcf98d004ed7779070c3da965c9
+oid sha256:846816e48b86e2be4cecdd5778e99f02cb73c4f53705b0b48f09130fb97960a4
 size 3087467144

train_results.json

@@ -1,9 +1,8 @@
 {
-    "epoch": 1.0,
-    "total_flos": 76916824473600.0,
-    "train_loss": 0.8026207946461333,
-    "train_runtime": 498.3301,
+    "total_flos": 76973799899136.0,
+    "train_loss": 0.8025032683942445,
+    "train_runtime": 710.0203,
     "train_samples": 16610,
-    "train_samples_per_second": 43.365,
-    "train_steps_per_second": 0.339
+    "train_samples_per_second": 30.454,
+    "train_steps_per_second": 0.238
 }

trainer_state.json

@@ -10,251 +10,286 @@
   "log_history": [
     {
       "epoch": 0.029585798816568046,
-      "grad_norm": 2.345319996716153,
+      "grad_norm": 2.2988141077652386,
       "learning_rate": 5.882352941176471e-06,
-      "loss": 1.0993,
+      "loss": 1.1002,
+      "mean_token_accuracy": 0.7101159904452009,
       "step": 5
     },
     {
       "epoch": 0.05917159763313609,
-      "grad_norm": 1.6336099031082492,
+      "grad_norm": 1.596218927081988,
       "learning_rate": 1.1764705882352942e-05,
-      "loss": 1.0402,
+      "loss": 1.0327,
+      "mean_token_accuracy": 0.720176724225474,
       "step": 10
     },
     {
       "epoch": 0.08875739644970414,
-      "grad_norm": 0.8859408831541707,
+      "grad_norm": 0.8479104246833407,
       "learning_rate": 1.7647058823529414e-05,
-      "loss": 0.9533,
+      "loss": 0.9517,
+      "mean_token_accuracy": 0.7325662989711975,
       "step": 15
     },
     {
       "epoch": 0.11834319526627218,
-      "grad_norm": 0.6064983041288236,
+      "grad_norm": 0.6465483847858254,
       "learning_rate": 1.9980782984658682e-05,
-      "loss": 0.8922,
+      "loss": 0.8804,
+      "mean_token_accuracy": 0.7473196725120462,
       "step": 20
     },
     {
       "epoch": 0.14792899408284024,
-      "grad_norm": 0.538397725140923,
+      "grad_norm": 0.5525325403311692,
       "learning_rate": 1.9863613034027224e-05,
-      "loss": 0.8552,
+      "loss": 0.8536,
+      "mean_token_accuracy": 0.7518004036197122,
       "step": 25
     },
     {
       "epoch": 0.17751479289940827,
-      "grad_norm": 0.4261073425530229,
+      "grad_norm": 0.44179450773629486,
       "learning_rate": 1.9641197940012136e-05,
-      "loss": 0.8283,
+      "loss": 0.8404,
+      "mean_token_accuracy": 0.7545674682549227,
       "step": 30
     },
     {
       "epoch": 0.20710059171597633,
-      "grad_norm": 0.37594749202019234,
+      "grad_norm": 0.39496081405164835,
       "learning_rate": 1.9315910880512792e-05,
-      "loss": 0.823,
+      "loss": 0.8212,
+      "mean_token_accuracy": 0.7580853602544403,
       "step": 35
     },
     {
       "epoch": 0.23668639053254437,
-      "grad_norm": 0.39908541118991764,
+      "grad_norm": 0.38859302520022476,
       "learning_rate": 1.8891222681391853e-05,
-      "loss": 0.8225,
+      "loss": 0.805,
+      "mean_token_accuracy": 0.7621072520755232,
       "step": 40
     },
     {
       "epoch": 0.26627218934911245,
-      "grad_norm": 0.3446073773136696,
+      "grad_norm": 0.4281961569580366,
       "learning_rate": 1.8371664782625287e-05,
-      "loss": 0.8073,
+      "loss": 0.7984,
+      "mean_token_accuracy": 0.7635130587409846,
       "step": 45
     },
     {
       "epoch": 0.2958579881656805,
-      "grad_norm": 0.33521545405571684,
+      "grad_norm": 0.393042490001915,
       "learning_rate": 1.7762780887657576e-05,
-      "loss": 0.7977,
+      "loss": 0.7936,
+      "mean_token_accuracy": 0.7642980542104124,
       "step": 50
     },
     {
       "epoch": 0.3254437869822485,
-      "grad_norm": 0.36639448217064385,
+      "grad_norm": 0.3719832470365729,
       "learning_rate": 1.7071067811865477e-05,
-      "loss": 0.7877,
+      "loss": 0.7894,
+      "mean_token_accuracy": 0.7649172633057038,
       "step": 55
     },
     {
       "epoch": 0.35502958579881655,
-      "grad_norm": 0.3696573153537768,
+      "grad_norm": 0.36967289202890546,
       "learning_rate": 1.6303906161279554e-05,
-      "loss": 0.7981,
+      "loss": 0.7999,
+      "mean_token_accuracy": 0.7619633834951797,
       "step": 60
     },
     {
       "epoch": 0.38461538461538464,
-      "grad_norm": 0.34317642014061395,
+      "grad_norm": 0.3965455034810465,
       "learning_rate": 1.5469481581224274e-05,
-      "loss": 0.7722,
+      "loss": 0.7867,
+      "mean_token_accuracy": 0.7647511953755232,
       "step": 65
     },
     {
       "epoch": 0.41420118343195267,
-      "grad_norm": 0.34403963076039407,
+      "grad_norm": 0.368068249126072,
       "learning_rate": 1.4576697415156818e-05,
-      "loss": 0.7741,
+      "loss": 0.7782,
+      "mean_token_accuracy": 0.7674480153107033,
       "step": 70
     },
     {
       "epoch": 0.4437869822485207,
-      "grad_norm": 0.33733303000440357,
+      "grad_norm": 0.3388950128403286,
       "learning_rate": 1.3635079705638298e-05,
-      "loss": 0.7852,
+      "loss": 0.777,
+      "mean_token_accuracy": 0.7675313738107395,
       "step": 75
     },
     {
       "epoch": 0.47337278106508873,
-      "grad_norm": 0.3519387066839729,
+      "grad_norm": 0.3430098886707366,
       "learning_rate": 1.2654675551080724e-05,
-      "loss": 0.7666,
+      "loss": 0.7783,
+      "mean_token_accuracy": 0.7666054602362825,
       "step": 80
     },
     {
       "epoch": 0.5029585798816568,
-      "grad_norm": 0.35934245892932387,
+      "grad_norm": 0.33478411969703326,
       "learning_rate": 1.164594590280734e-05,
-      "loss": 0.7706,
+      "loss": 0.7785,
+      "mean_token_accuracy": 0.7669639865984688,
       "step": 85
     },
     {
       "epoch": 0.5325443786982249,
-      "grad_norm": 0.3600712658909069,
+      "grad_norm": 0.34768462687825924,
       "learning_rate": 1.0619653946285948e-05,
-      "loss": 0.7699,
+      "loss": 0.7723,
+      "mean_token_accuracy": 0.7682154932141926,
       "step": 90
     },
     {
       "epoch": 0.5621301775147929,
-      "grad_norm": 0.3623401906293079,
+      "grad_norm": 0.3445315216411646,
       "learning_rate": 9.586750257511868e-06,
-      "loss": 0.7771,
+      "loss": 0.7668,
+      "mean_token_accuracy": 0.7698424375606829,
       "step": 95
     },
     {
       "epoch": 0.591715976331361,
-      "grad_norm": 0.3321501796878117,
+      "grad_norm": 0.3412337472975462,
       "learning_rate": 8.558255959926533e-06,
-      "loss": 0.753,
+      "loss": 0.7582,
+      "mean_token_accuracy": 0.7714304347299908,
       "step": 100
     },
     {
       "epoch": 0.591715976331361,
-      "eval_loss": 0.7878655195236206,
-      "eval_runtime": 0.782,
-      "eval_samples_per_second": 163.676,
-      "eval_steps_per_second": 5.115,
+      "eval_loss": 0.7885846495628357,
+      "eval_mean_token_accuracy": 0.7482494053174686,
+      "eval_runtime": 2.2356,
+      "eval_samples_per_second": 57.703,
+      "eval_steps_per_second": 2.237,
       "step": 100
     },
     {
       "epoch": 0.621301775147929,
-      "grad_norm": 0.3343545042291716,
+      "grad_norm": 0.35299455458895956,
       "learning_rate": 7.545145128592009e-06,
-      "loss": 0.7569,
+      "loss": 0.759,
+      "mean_token_accuracy": 0.7714738320259433,
       "step": 105
     },
     {
       "epoch": 0.650887573964497,
-      "grad_norm": 0.3444014813426581,
+      "grad_norm": 0.31616289960261323,
       "learning_rate": 6.558227696373617e-06,
-      "loss": 0.768,
+      "loss": 0.7775,
+      "mean_token_accuracy": 0.7664820794613972,
       "step": 110
     },
     {
       "epoch": 0.6804733727810651,
-      "grad_norm": 0.3079534620667556,
+      "grad_norm": 0.307888028457379,
       "learning_rate": 5.608034111526298e-06,
-      "loss": 0.7623,
+      "loss": 0.7636,
+      "mean_token_accuracy": 0.7704197310032757,
       "step": 115
     },
     {
       "epoch": 0.7100591715976331,
-      "grad_norm": 0.2980052178831321,
+      "grad_norm": 0.2995694223287612,
       "learning_rate": 4.704702977392914e-06,
-      "loss": 0.7513,
+      "loss": 0.7511,
+      "mean_token_accuracy": 0.7742426406160382,
       "step": 120
     },
     {
       "epoch": 0.7396449704142012,
-      "grad_norm": 0.3422464501794574,
+      "grad_norm": 0.31585565817231076,
       "learning_rate": 3.857872873103322e-06,
-      "loss": 0.7537,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7717561342325581,
       "step": 125
     },
     {
       "epoch": 0.7692307692307693,
-      "grad_norm": 0.31159482318154197,
+      "grad_norm": 0.32069400376048846,
       "learning_rate": 3.0765795095517026e-06,
-      "loss": 0.7555,
+      "loss": 0.7624,
+      "mean_token_accuracy": 0.7705784584578896,
       "step": 130
     },
     {
       "epoch": 0.7988165680473372,
-      "grad_norm": 0.3216694034093574,
+      "grad_norm": 0.30337427328877,
       "learning_rate": 2.369159318001937e-06,
-      "loss": 0.7583,
+      "loss": 0.7684,
+      "mean_token_accuracy": 0.7688614612364428,
       "step": 135
     },
     {
       "epoch": 0.8284023668639053,
-      "grad_norm": 0.29895838755514975,
+      "grad_norm": 0.2931589273225087,
       "learning_rate": 1.743160500034443e-06,
-      "loss": 0.7498,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.7717145410630937,
       "step": 140
     },
     {
       "epoch": 0.8579881656804734,
-      "grad_norm": 0.3121733796733359,
+      "grad_norm": 0.31239776071011455,
       "learning_rate": 1.2052624879351105e-06,
-      "loss": 0.7565,
+      "loss": 0.7595,
+      "mean_token_accuracy": 0.7714656640175818,
       "step": 145
     },
     {
       "epoch": 0.8875739644970414,
-      "grad_norm": 0.29445775165969484,
+      "grad_norm": 0.2909480842066185,
       "learning_rate": 7.612046748871327e-07,
-      "loss": 0.7665,
+      "loss": 0.7505,
+      "mean_token_accuracy": 0.7739669991508189,
       "step": 150
     },
     {
       "epoch": 0.9171597633136095,
-      "grad_norm": 0.2963830850309388,
+      "grad_norm": 0.279458821658001,
       "learning_rate": 4.1572517541747294e-07,
-      "loss": 0.7613,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.7767323569336231,
       "step": 155
     },
     {
       "epoch": 0.9467455621301775,
-      "grad_norm": 0.2863988168086613,
+      "grad_norm": 0.29648687488650516,
       "learning_rate": 1.7251026952640583e-07,
-      "loss": 0.7606,
+      "loss": 0.754,
+      "mean_token_accuracy": 0.7732043043845805,
       "step": 160
     },
     {
       "epoch": 0.9763313609467456,
-      "grad_norm": 0.2760424898927774,
+      "grad_norm": 0.2773995248389471,
       "learning_rate": 3.4155069933301535e-08,
-      "loss": 0.7445,
+      "loss": 0.7467,
+      "mean_token_accuracy": 0.7749679952859235,
       "step": 165
     },
     {
       "epoch": 1.0,
+      "mean_token_accuracy": 0.7705127247007673,
       "step": 169,
-      "total_flos": 76916824473600.0,
-      "train_loss": 0.8026207946461333,
-      "train_runtime": 498.3301,
-      "train_samples_per_second": 43.365,
-      "train_steps_per_second": 0.339
+      "total_flos": 76973799899136.0,
+      "train_loss": 0.8025032683942445,
+      "train_runtime": 710.0203,
+      "train_samples_per_second": 30.454,
+      "train_steps_per_second": 0.238
     }
   ],
   "logging_steps": 5,
@@ -274,8 +309,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 76916824473600.0,
-  "train_batch_size": 4,
+  "total_flos": 76973799899136.0,
+  "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:237a15eca0faf2c43d81e67abeea627888a117b938f040d9bf15c055f4cd8fe2
-size 7352
+oid sha256:5e7043d0dcec036604186a342c05b4b1b9b359afc64da2f56ce61db35f0a4ae1
+size 7416