End of training

README.md

@@ -3,9 +3,16 @@ library_name: transformers
 license: llama3.2
 base_model: tanliboy/llama-3.2-3b-sft-2
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- HuggingFaceH4/orca_dpo_pairs
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: llama-3.2-3b-dpo-2
   results: []
@@ -16,17 +23,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # llama-3.2-3b-dpo-2
 
-This model is a fine-tuned version of [tanliboy/llama-3.2-3b-sft-2](https://huggingface.co/tanliboy/llama-3.2-3b-sft-2) on the None dataset.
+This model is a fine-tuned version of [tanliboy/llama-3.2-3b-sft-2](https://huggingface.co/tanliboy/llama-3.2-3b-sft-2) on the HuggingFaceH4/orca_dpo_pairs and the HuggingFaceH4/ultrafeedback_binarized datasets.
 It achieves the following results on the evaluation set:
-- Loss: 0.5808
-- Rewards/chosen: 1.8125
-- Rewards/rejected: -4.0822
-- Rewards/accuracies: 0.7880
-- Rewards/margins: 5.8947
-- Logps/rejected: -387.3112
-- Logps/chosen: -337.8669
-- Logits/rejected: 0.2355
-- Logits/chosen: 0.1785
+- Loss: 0.5814
+- Rewards/chosen: 1.7432
+- Rewards/rejected: -4.1735
+- Rewards/accuracies: 0.7848
+- Rewards/margins: 5.9167
+- Logps/rejected: -388.2242
+- Logps/chosen: -338.5596
+- Logits/rejected: 0.2395
+- Logits/chosen: 0.1826
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 2.998693948628646,
+    "eval_logits/chosen": 0.182608962059021,
+    "eval_logits/rejected": 0.23953425884246826,
+    "eval_logps/chosen": -338.5595703125,
+    "eval_logps/rejected": -388.22418212890625,
+    "eval_loss": 0.5813568234443665,
+    "eval_rewards/accuracies": 0.7848101258277893,
+    "eval_rewards/chosen": 1.7432384490966797,
+    "eval_rewards/margins": 5.916741847991943,
+    "eval_rewards/rejected": -4.173503398895264,
+    "eval_runtime": 69.9912,
+    "eval_samples": 2500,
+    "eval_samples_per_second": 35.719,
+    "eval_steps_per_second": 1.129,
     "total_flos": 0.0,
     "train_loss": 0.5855818307081304,
     "train_runtime": 16735.6732,

config.json

@@ -35,6 +35,6 @@
   "tie_word_embeddings": true,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.44.2",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 128256
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 2.998693948628646,
+    "eval_logits/chosen": 0.182608962059021,
+    "eval_logits/rejected": 0.23953425884246826,
+    "eval_logps/chosen": -338.5595703125,
+    "eval_logps/rejected": -388.22418212890625,
+    "eval_loss": 0.5813568234443665,
+    "eval_rewards/accuracies": 0.7848101258277893,
+    "eval_rewards/chosen": 1.7432384490966797,
+    "eval_rewards/margins": 5.916741847991943,
+    "eval_rewards/rejected": -4.173503398895264,
+    "eval_runtime": 69.9912,
+    "eval_samples": 2500,
+    "eval_samples_per_second": 35.719,
+    "eval_steps_per_second": 1.129
+}

runs/Oct01_07-05-14_action-graph-trainer/events.out.tfevents.1727784078.action-graph-trainer.1627413.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8121598fcd5f3d6784b478a733c0e808739d2853cba608136ef552ae0accc7ff
+size 828