Model save

README.md

@@ -1,15 +1,11 @@
 ---
-license: apache-2.0
+license: mit
 library_name: peft
 tags:
-- alignment-handbook
-- generated_from_trainer
 - trl
 - dpo
 - generated_from_trainer
 base_model: DUAL-GPO/phi-2-gpo-new-i0
-datasets:
-- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: phi-2-gpo-v6-i1
   results: []
@@ -20,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # phi-2-gpo-v6-i1
 
-This model is a fine-tuned version of [DUAL-GPO/phi-2-gpo-new-i0](https://huggingface.co/DUAL-GPO/phi-2-gpo-new-i0) on the HuggingFaceH4/ultrafeedback_binarized dataset.
+This model is a fine-tuned version of [DUAL-GPO/phi-2-gpo-new-i0](https://huggingface.co/DUAL-GPO/phi-2-gpo-new-i0) on the None dataset.
 
 ## Model description
 
@@ -44,12 +40,14 @@ The following hyperparameters were used during training:
 - eval_batch_size: 4
 - seed: 42
 - distributed_type: multi-GPU
+- num_devices: 3
 - gradient_accumulation_steps: 4
-- total_train_batch_size: 16
+- total_train_batch_size: 48
+- total_eval_batch_size: 12
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 1
+- num_epochs: 2
 
 ### Training results
 

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b7af97ea380712e402312cd49eff36da55468c74765e36c98f52e0e0ac294dff
+oid sha256:7b259973d90c50e4d7da05879aa5daf8af4b2cb88c49cc1ee7d4b375d3b074cd
 size 167807296

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.0,
-    "train_loss": 0.24134081795175627,
-    "train_runtime": 12108.13,
-    "train_samples": 21000,
-    "train_samples_per_second": 1.734,
-    "train_steps_per_second": 0.108
+    "epoch": 2.0,
+    "train_loss": 0.27172684411589915,
+    "train_runtime": 11567.6763,
+    "train_samples": 20000,
+    "train_samples_per_second": 3.458,
+    "train_steps_per_second": 0.072
 }

runs/May15_15-27-49_gpu4-119-5/events.out.tfevents.1715750995.gpu4-119-5.3097660.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cfcbe207bb9600ba418e6086d5a50b60651ecde83e206b41bacdc355db7289cc
-size 56037
+oid sha256:63307d23b36188e99c93745be59a64b5614b33715f51682ad83e6f075aeefd92
+size 58293

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.0,
-    "train_loss": 0.24134081795175627,
-    "train_runtime": 12108.13,
-    "train_samples": 21000,
-    "train_samples_per_second": 1.734,
-    "train_steps_per_second": 0.108
+    "epoch": 2.0,
+    "train_loss": 0.27172684411589915,
+    "train_runtime": 11567.6763,
+    "train_samples": 20000,
+    "train_samples_per_second": 3.458,
+    "train_steps_per_second": 0.072
 }