Model save

README.md

@@ -2,15 +2,10 @@
 license: llama2
 base_model: Minbyul/meditron-7b-wo-medication_qa-sft
 tags:
-- alignment-handbook
-- trl
-- dpo
-- generated_from_trainer
 - trl
 - dpo
+- alignment-handbook
 - generated_from_trainer
-datasets:
-- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: meditron-7b-dpo-full-sft-wo-medication_qa
   results: []
@@ -21,17 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # meditron-7b-dpo-full-sft-wo-medication_qa
 
-This model is a fine-tuned version of [Minbyul/meditron-7b-wo-medication_qa-sft](https://huggingface.co/Minbyul/meditron-7b-wo-medication_qa-sft) on the HuggingFaceH4/ultrafeedback_binarized dataset.
-It achieves the following results on the evaluation set:
-- Loss: 0.6342
-- Rewards/chosen: -0.0725
-- Rewards/rejected: -0.3037
-- Rewards/accuracies: 0.7812
-- Rewards/margins: 0.2312
-- Logps/rejected: -613.1050
-- Logps/chosen: -394.9411
-- Logits/rejected: -1.0634
-- Logits/chosen: -1.1747
+This model is a fine-tuned version of [Minbyul/meditron-7b-wo-medication_qa-sft](https://huggingface.co/Minbyul/meditron-7b-wo-medication_qa-sft) on an unknown dataset.
 
 ## Model description
 

all_results.json

@@ -13,9 +13,9 @@
     "eval_samples": 240,
     "eval_samples_per_second": 7.857,
     "eval_steps_per_second": 0.262,
-    "train_loss": 0.5082056158680027,
-    "train_runtime": 904.9767,
+    "train_loss": 0.5081947011462713,
+    "train_runtime": 896.7003,
     "train_samples": 3790,
-    "train_samples_per_second": 4.188,
-    "train_steps_per_second": 0.065
+    "train_samples_per_second": 4.227,
+    "train_steps_per_second": 0.066
 }

config.json

@@ -23,6 +23,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.39.0.dev0",
-  "use_cache": true,
+  "use_cache": false,
   "vocab_size": 32017
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a219f5b9727cc10c912ebb17f51e961332273fb6d58f0bec7971ffc3a20372a7
+oid sha256:62ddb60864ab3df63427dd2ef97cefbe278cddbd8ef4b58ab3373cbe51997f4b
 size 4939124616

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:846a04c4147ee488539d490649c404aa373290076c967d8a371dca5c77e71c5e
+oid sha256:451f3f655b623958dd20a8428b6ec7f29f95ff71e5ab20b43810ce0c55a5e275
 size 4947390880

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5aa02fc6a945bd66fa64f158ec7da4a6ba11ba3803cb1e6f10710a05d0d93206
+oid sha256:8b1bbd45a4105bf147ff5fb16fd4d7a579b997486765402e072d695babddf2ce
 size 3590628080

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 0.99,
-    "train_loss": 0.5082056158680027,
-    "train_runtime": 904.9767,
+    "train_loss": 0.5081947011462713,
+    "train_runtime": 896.7003,
     "train_samples": 3790,
-    "train_samples_per_second": 4.188,
-    "train_steps_per_second": 0.065
+    "train_samples_per_second": 4.227,
+    "train_steps_per_second": 0.066
 }

trainer_state.json

@@ -10,7 +10,7 @@
   "log_history": [
     {
       "epoch": 0.02,
-      "grad_norm": 6.891863950600918,
+      "grad_norm": 6.892402047143212,
       "learning_rate": 8.333333333333333e-08,
       "logits/chosen": -1.1214768886566162,
       "logits/rejected": -1.0666239261627197,
@@ -25,87 +25,87 @@
     },
     {
       "epoch": 0.17,
-      "grad_norm": 6.6410695205888075,
+      "grad_norm": 6.623544286492592,
       "learning_rate": 4.930057285201027e-07,
-      "logits/chosen": -1.2806458473205566,
-      "logits/rejected": -0.9247087836265564,
-      "logps/chosen": -503.0013122558594,
-      "logps/rejected": -899.8956298828125,
-      "loss": 0.6907,
-      "rewards/accuracies": 0.5069444179534912,
-      "rewards/chosen": -0.0002252649428555742,
-      "rewards/margins": 0.004411212634295225,
-      "rewards/rejected": -0.004636477679014206,
+      "logits/chosen": -1.280522346496582,
+      "logits/rejected": -0.9248583912849426,
+      "logps/chosen": -503.0466613769531,
+      "logps/rejected": -899.9366455078125,
+      "loss": 0.6904,
+      "rewards/accuracies": 0.5347222089767456,
+      "rewards/chosen": -0.0006784469587728381,
+      "rewards/margins": 0.004367371555417776,
+      "rewards/rejected": -0.005045818164944649,
       "step": 10
     },
     {
       "epoch": 0.34,
-      "grad_norm": 7.3762614729181015,
+      "grad_norm": 7.366007707770025,
       "learning_rate": 4.187457503795526e-07,
-      "logits/chosen": -1.2530821561813354,
-      "logits/rejected": -0.9845311045646667,
-      "logps/chosen": -543.0537109375,
-      "logps/rejected": -933.8472900390625,
+      "logits/chosen": -1.253061294555664,
+      "logits/rejected": -0.9847872853279114,
+      "logps/chosen": -543.08447265625,
+      "logps/rejected": -933.7767333984375,
       "loss": 0.6527,
-      "rewards/accuracies": 0.8687499761581421,
-      "rewards/chosen": -0.019722815603017807,
-      "rewards/margins": 0.08204329013824463,
-      "rewards/rejected": -0.10176609456539154,
+      "rewards/accuracies": 0.856249988079071,
+      "rewards/chosen": -0.020030761137604713,
+      "rewards/margins": 0.08102954924106598,
+      "rewards/rejected": -0.10106030851602554,
       "step": 20
     },
     {
       "epoch": 0.5,
-      "grad_norm": 8.929333715687283,
+      "grad_norm": 8.97929873252708,
       "learning_rate": 2.8691164100062034e-07,
-      "logits/chosen": -1.1901506185531616,
-      "logits/rejected": -0.9701334238052368,
-      "logps/chosen": -547.3389892578125,
-      "logps/rejected": -972.1591796875,
-      "loss": 0.5279,
-      "rewards/accuracies": 0.8812500238418579,
-      "rewards/chosen": -0.08818452060222626,
-      "rewards/margins": 0.44495606422424316,
-      "rewards/rejected": -0.5331405997276306,
+      "logits/chosen": -1.188706398010254,
+      "logits/rejected": -0.9695678949356079,
+      "logps/chosen": -547.2942504882812,
+      "logps/rejected": -972.0201416015625,
+      "loss": 0.5278,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": -0.08773749321699142,
+      "rewards/margins": 0.4440121054649353,
+      "rewards/rejected": -0.5317496061325073,
       "step": 30
     },
     {
       "epoch": 0.67,
-      "grad_norm": 7.370675587057317,
+      "grad_norm": 7.373164759260948,
       "learning_rate": 1.4248369943086995e-07,
-      "logits/chosen": -1.121147632598877,
-      "logits/rejected": -0.9864951372146606,
-      "logps/chosen": -544.5201416015625,
-      "logps/rejected": -1050.002197265625,
+      "logits/chosen": -1.121267557144165,
+      "logits/rejected": -0.9862432479858398,
+      "logps/chosen": -544.4817504882812,
+      "logps/rejected": -1049.937744140625,
       "loss": 0.4145,
-      "rewards/accuracies": 0.90625,
-      "rewards/chosen": -0.19094732403755188,
-      "rewards/margins": 0.9460043907165527,
-      "rewards/rejected": -1.1369515657424927,
+      "rewards/accuracies": 0.8999999761581421,
+      "rewards/chosen": -0.19056306779384613,
+      "rewards/margins": 0.9457426071166992,
+      "rewards/rejected": -1.136305570602417,
       "step": 40
     },
     {
       "epoch": 0.84,
-      "grad_norm": 6.641314363386866,
+      "grad_norm": 6.631179730087488,
       "learning_rate": 3.473909705816111e-08,
-      "logits/chosen": -1.0754799842834473,
-      "logits/rejected": -1.0215175151824951,
-      "logps/chosen": -568.5784912109375,
-      "logps/rejected": -1094.7982177734375,
+      "logits/chosen": -1.0745445489883423,
+      "logits/rejected": -1.0211777687072754,
+      "logps/chosen": -568.6431884765625,
+      "logps/rejected": -1094.7789306640625,
       "loss": 0.3982,
-      "rewards/accuracies": 0.875,
-      "rewards/chosen": -0.3303142786026001,
-      "rewards/margins": 1.3637864589691162,
-      "rewards/rejected": -1.6941007375717163,
+      "rewards/accuracies": 0.8687499761581421,
+      "rewards/chosen": -0.33096131682395935,
+      "rewards/margins": 1.3629460334777832,
+      "rewards/rejected": -1.6939074993133545,
       "step": 50
     },
     {
       "epoch": 0.99,
       "step": 59,
       "total_flos": 0.0,
-      "train_loss": 0.5082056158680027,
-      "train_runtime": 904.9767,
-      "train_samples_per_second": 4.188,
-      "train_steps_per_second": 0.065
+      "train_loss": 0.5081947011462713,
+      "train_runtime": 896.7003,
+      "train_samples_per_second": 4.227,
+      "train_steps_per_second": 0.066
     }
   ],
   "logging_steps": 10,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:20c2e7d7bca5ee6f2a1963f9a3c804561f4d02a9ab170c8917e179691a8b2c30
+oid sha256:6e8ba9506cb4ca202b7c2f4ebbae99b55462a4f2328c8eec10bf0c8002349ade
 size 6264