End of training

README.md

@@ -2,9 +2,15 @@
 license: apache-2.0
 base_model: martimfasantos/tinyllama-1.1b-sum-sft-full
 tags:
+- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
+- trl
+- dpo
+- generated_from_trainer
+datasets:
+- openai/summarize_from_feedback
 model-index:
 - name: tinyllama-1.1b-sum-dpo-full_LR5e-7_3epochs
   results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # tinyllama-1.1b-sum-dpo-full_LR5e-7_3epochs
 
-This model is a fine-tuned version of [martimfasantos/tinyllama-1.1b-sum-sft-full](https://huggingface.co/martimfasantos/tinyllama-1.1b-sum-sft-full) on an unknown dataset.
+This model is a fine-tuned version of [martimfasantos/tinyllama-1.1b-sum-sft-full](https://huggingface.co/martimfasantos/tinyllama-1.1b-sum-sft-full) on the openai/summarize_from_feedback dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.7095
-- Rewards/chosen: -2.8594
-- Rewards/rejected: -3.4153
-- Rewards/accuracies: 0.6336
-- Rewards/margins: 0.5559
-- Logps/rejected: -404.2798
-- Logps/chosen: -344.9560
-- Logits/rejected: -1.9830
-- Logits/chosen: -2.0075
+- Loss: 0.7099
+- Rewards/chosen: -2.8601
+- Rewards/rejected: -3.4154
+- Rewards/accuracies: 0.6320
+- Rewards/margins: 0.5553
+- Logps/rejected: -404.2897
+- Logps/chosen: -345.0273
+- Logits/rejected: -1.9822
+- Logits/chosen: -2.0068
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 3.0,
+    "eval_logits/chosen": -2.0068085193634033,
+    "eval_logits/rejected": -1.9822146892547607,
+    "eval_logps/chosen": -345.02728271484375,
+    "eval_logps/rejected": -404.28973388671875,
+    "eval_loss": 0.7099329233169556,
+    "eval_rewards/accuracies": 0.6319702863693237,
+    "eval_rewards/chosen": -2.8601176738739014,
+    "eval_rewards/margins": 0.5552833080291748,
+    "eval_rewards/rejected": -3.4154014587402344,
+    "eval_runtime": 360.569,
+    "eval_samples": 4304,
+    "eval_samples_per_second": 11.937,
+    "eval_steps_per_second": 1.492,
     "total_flos": 0.0,
     "train_loss": 0.49131362667692796,
     "train_runtime": 84120.8122,

config.json

@@ -24,6 +24,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "float32",
   "transformers_version": "4.41.2",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }

eval_results.json

@@ -0,0 +1,16 @@
+{
+    "epoch": 3.0,
+    "eval_logits/chosen": -2.0068085193634033,
+    "eval_logits/rejected": -1.9822146892547607,
+    "eval_logps/chosen": -345.02728271484375,
+    "eval_logps/rejected": -404.28973388671875,
+    "eval_loss": 0.7099329233169556,
+    "eval_rewards/accuracies": 0.6319702863693237,
+    "eval_rewards/chosen": -2.8601176738739014,
+    "eval_rewards/margins": 0.5552833080291748,
+    "eval_rewards/rejected": -3.4154014587402344,
+    "eval_runtime": 360.569,
+    "eval_samples": 4304,
+    "eval_samples_per_second": 11.937,
+    "eval_steps_per_second": 1.492
+}

runs/Jun07_09-08-56_poseidon/events.out.tfevents.1717836367.poseidon.3198658.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b1d1167a242fd59d22039df9dcca3d4ec15190211b5658f5938e55a524c52c4
+size 841