End of training

README.md

@@ -2,9 +2,15 @@
 license: apache-2.0
 base_model: martimfasantos/tinyllama-1.1b-sum-sft-full_old
 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-8_BS64_3epochs_old
   results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # tinyllama-1.1b-sum-dpo-full_LR5e-8_BS64_3epochs_old
 
-This model is a fine-tuned version of [martimfasantos/tinyllama-1.1b-sum-sft-full_old](https://huggingface.co/martimfasantos/tinyllama-1.1b-sum-sft-full_old) on an unknown dataset.
+This model is a fine-tuned version of [martimfasantos/tinyllama-1.1b-sum-sft-full_old](https://huggingface.co/martimfasantos/tinyllama-1.1b-sum-sft-full_old) on the openai/summarize_from_feedback dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.6852
-- Rewards/chosen: -0.0659
-- Rewards/rejected: -0.0838
-- Rewards/accuracies: 0.6006
+- Loss: 0.6851
+- Rewards/chosen: -0.0660
+- Rewards/rejected: -0.0839
+- Rewards/accuracies: 0.5978
 - Rewards/margins: 0.0179
-- Logps/rejected: -71.5612
-- Logps/chosen: -65.3069
-- Logits/rejected: -3.0327
-- Logits/chosen: -3.0385
+- Logps/rejected: -71.5685
+- Logps/chosen: -65.3140
+- Logits/rejected: -3.0328
+- Logits/chosen: -3.0386
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 3.0,
+    "eval_logits/chosen": -3.0385637283325195,
+    "eval_logits/rejected": -3.0328078269958496,
+    "eval_logps/chosen": -65.31404876708984,
+    "eval_logps/rejected": -71.56845092773438,
+    "eval_loss": 0.6851488351821899,
+    "eval_rewards/accuracies": 0.597815990447998,
+    "eval_rewards/chosen": -0.06602144241333008,
+    "eval_rewards/margins": 0.01786196418106556,
+    "eval_rewards/rejected": -0.08388339728116989,
+    "eval_runtime": 382.154,
+    "eval_samples": 4304,
+    "eval_samples_per_second": 11.262,
+    "eval_steps_per_second": 1.408,
     "total_flos": 0.0,
     "train_loss": 0.6826107928102766,
     "train_runtime": 83562.7786,

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": -3.0385637283325195,
+    "eval_logits/rejected": -3.0328078269958496,
+    "eval_logps/chosen": -65.31404876708984,
+    "eval_logps/rejected": -71.56845092773438,
+    "eval_loss": 0.6851488351821899,
+    "eval_rewards/accuracies": 0.597815990447998,
+    "eval_rewards/chosen": -0.06602144241333008,
+    "eval_rewards/margins": 0.01786196418106556,
+    "eval_rewards/rejected": -0.08388339728116989,
+    "eval_runtime": 382.154,
+    "eval_samples": 4304,
+    "eval_samples_per_second": 11.262,
+    "eval_steps_per_second": 1.408
+}

runs/Jun27_22-38-11_poseidon/events.out.tfevents.1719612340.poseidon.708792.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3aff2620ca0391cab8ebe8b01694d13f59842a38954a61e20ef05840fa0e1d13
+size 828