End of training

README.md

@@ -1,9 +1,16 @@
 ---
 library_name: transformers
+base_model: data/OpenELM-1_1B-SFT
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: OpenELM-1_1B-DPO-2
   results: []
@@ -14,7 +21,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # OpenELM-1_1B-DPO-2
 
-This model was trained from scratch on an unknown dataset.
+This model is a fine-tuned version of [data/OpenELM-1_1B-SFT](https://huggingface.co/data/OpenELM-1_1B-SFT) on the HuggingFaceH4/ultrafeedback_binarized dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.7793
 - Rewards/chosen: -11.75

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 2.0,
+    "eval_logits/chosen": -13.3125,
+    "eval_logits/rejected": -13.0625,
+    "eval_logps/chosen": -556.0,
+    "eval_logps/rejected": -564.0,
+    "eval_loss": 0.7793359160423279,
+    "eval_rewards/accuracies": 0.72265625,
+    "eval_rewards/chosen": -11.75,
+    "eval_rewards/margins": 1.9140625,
+    "eval_rewards/rejected": -13.6875,
+    "eval_runtime": 46.6612,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 42.862,
+    "eval_steps_per_second": 0.686,
     "total_flos": 0.0,
     "train_loss": 0.41369995401518184,
     "train_runtime": 7266.4225,

config.json

@@ -120,6 +120,6 @@
   "share_input_output_layers": true,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.44.2",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 2.0,
+    "eval_logits/chosen": -13.3125,
+    "eval_logits/rejected": -13.0625,
+    "eval_logps/chosen": -556.0,
+    "eval_logps/rejected": -564.0,
+    "eval_loss": 0.7793359160423279,
+    "eval_rewards/accuracies": 0.72265625,
+    "eval_rewards/chosen": -11.75,
+    "eval_rewards/margins": 1.9140625,
+    "eval_rewards/rejected": -13.6875,
+    "eval_runtime": 46.6612,
+    "eval_samples": 2000,
+    "eval_samples_per_second": 42.862,
+    "eval_steps_per_second": 0.686
+}

runs/Aug27_17-49-13_xe8545-a100-16/events.out.tfevents.1724781612.xe8545-a100-16.2588569.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1eef41806eeecd799871db59c654ecf90ae29ebfa40db48fd020b44026bf09ed
+size 828