Model save

README.md

@@ -2,16 +2,10 @@
 license: mit
 base_model: HuggingFaceH4/mistral-7b-sft-beta
 tags:
-- alignment-handbook
-- trl
-- dpo
-- generated_from_trainer
 - trl
 - dpo
 - alignment-handbook
 - generated_from_trainer
-datasets:
-- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: DPO-Zephyr-7B
   results: []
@@ -22,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # DPO-Zephyr-7B
 
-This model is a fine-tuned version of [HuggingFaceH4/mistral-7b-sft-beta](https://huggingface.co/HuggingFaceH4/mistral-7b-sft-beta) on the HuggingFaceH4/ultrafeedback_binarized dataset.
+This model is a fine-tuned version of [HuggingFaceH4/mistral-7b-sft-beta](https://huggingface.co/HuggingFaceH4/mistral-7b-sft-beta) on the None dataset.
 
 ## Model description
 
@@ -47,8 +41,8 @@ The following hyperparameters were used during training:
 - seed: 42
 - distributed_type: multi-GPU
 - num_devices: 8
-- gradient_accumulation_steps: 4
-- total_train_batch_size: 128
+- gradient_accumulation_steps: 8
+- total_train_batch_size: 256
 - total_eval_batch_size: 64
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine

all_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 0.99581589958159,
+    "epoch": 0.8368200836820083,
     "total_flos": 0.0,
-    "train_loss": 0.09238386955581793,
-    "train_runtime": 386.322,
+    "train_loss": 0.0,
+    "train_runtime": 172.1278,
     "train_samples": 15283,
-    "train_samples_per_second": 39.56,
-    "train_steps_per_second": 0.308
+    "train_samples_per_second": 88.789,
+    "train_steps_per_second": 0.343
 }

config.json

@@ -21,6 +21,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.40.2",
-  "use_cache": true,
+  "use_cache": false,
   "vocab_size": 32000
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:af488688f56cda56ff87ad1eec035fa9ae3a85bca93bdbe59211aa9ae3eae187
+oid sha256:ed828d1d3747d787beccc5e05e82a27cf053394733010c14dce37af5e1b14050
 size 4943162336

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3768b0b7c81899c59e9afb844018c1b1f295d77c8c47e5c118e3562a776ae5a7
+oid sha256:3e1be7f1462cc851f1af5cb4395cd8721ff70220bf4f51337337e3d64ef195b4
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:16f3d87f94195cc6683db7aeafd3e916dad7e931f54a05872d55a851ad2d0ed0
+oid sha256:65324868754ad5cbed72f6dce5c6497a4e143e70c2c101b69cdd69df2e3d31cf
 size 4540516344

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 0.99581589958159,
+    "epoch": 0.8368200836820083,
     "total_flos": 0.0,
-    "train_loss": 0.09238386955581793,
-    "train_runtime": 386.322,
+    "train_loss": 0.0,
+    "train_runtime": 172.1278,
     "train_samples": 15283,
-    "train_samples_per_second": 39.56,
-    "train_steps_per_second": 0.308
+    "train_samples_per_second": 88.789,
+    "train_steps_per_second": 0.343
 }

trainer_state.json

@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.99581589958159,
+  "epoch": 0.8368200836820083,
   "eval_steps": 500,
-  "global_step": 119,
+  "global_step": 100,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -174,32 +174,17 @@
       "step": 100
     },
     {
-      "epoch": 0.9205020920502092,
-      "grad_norm": 15.746470079783288,
-      "learning_rate": 8.677580722139671e-09,
-      "logits/chosen": -2.7727127075195312,
-      "logits/rejected": -2.7130322456359863,
-      "logps/chosen": -367.272216796875,
-      "logps/rejected": -343.4369201660156,
-      "loss": 0.5945,
-      "rewards/accuracies": 0.699999988079071,
-      "rewards/chosen": -0.46789541840553284,
-      "rewards/margins": 0.3372054696083069,
-      "rewards/rejected": -0.8051007986068726,
-      "step": 110
-    },
-    {
-      "epoch": 0.99581589958159,
-      "step": 119,
+      "epoch": 0.8368200836820083,
+      "step": 100,
       "total_flos": 0.0,
-      "train_loss": 0.09238386955581793,
-      "train_runtime": 386.322,
-      "train_samples_per_second": 39.56,
-      "train_steps_per_second": 0.308
+      "train_loss": 0.0,
+      "train_runtime": 172.1278,
+      "train_samples_per_second": 88.789,
+      "train_steps_per_second": 0.343
     }
   ],
   "logging_steps": 10,
-  "max_steps": 119,
+  "max_steps": 59,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:128ab99ba4e1911f9502d9e9cba3a105bb6a758aa83704d9f3e92a77614e5773
+oid sha256:26d85afdfb7f05f00d4e8812ab20799311b4401986345249d4734fdf1972cd0b
 size 6264