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_LR1e-7_3epochs
   results: []
@@ -15,17 +21,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # tinyllama-1.1b-sum-dpo-full_LR1e-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: 0.6500
-- Rewards/chosen: -1.0595
-- Rewards/rejected: -1.2334
-- Rewards/accuracies: 0.6046
+- Loss: 0.6501
+- Rewards/chosen: -1.0591
+- Rewards/rejected: -1.2329
+- Rewards/accuracies: 0.6032
 - Rewards/margins: 0.1739
-- Logps/rejected: -186.0905
-- Logps/chosen: -164.9614
-- Logits/rejected: -2.3429
-- Logits/chosen: -2.3549
+- Logps/rejected: -186.0431
+- Logps/chosen: -164.9210
+- Logits/rejected: -2.3430
+- Logits/chosen: -2.3551
 
 ## Model description
 

all_results.json

@@ -1,5 +1,18 @@
 {
     "epoch": 3.0,
+    "eval_logits/chosen": -2.355067729949951,
+    "eval_logits/rejected": -2.3430099487304688,
+    "eval_logps/chosen": -164.92103576660156,
+    "eval_logps/rejected": -186.04312133789062,
+    "eval_loss": 0.6500672101974487,
+    "eval_rewards/accuracies": 0.6031598448753357,
+    "eval_rewards/chosen": -1.0590553283691406,
+    "eval_rewards/margins": 0.1738797128200531,
+    "eval_rewards/rejected": -1.2329350709915161,
+    "eval_runtime": 384.1905,
+    "eval_samples": 4304,
+    "eval_samples_per_second": 11.203,
+    "eval_steps_per_second": 1.4,
     "total_flos": 0.0,
     "train_loss": 0.6300815686244623,
     "train_runtime": 86182.1351,

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": -2.355067729949951,
+    "eval_logits/rejected": -2.3430099487304688,
+    "eval_logps/chosen": -164.92103576660156,
+    "eval_logps/rejected": -186.04312133789062,
+    "eval_loss": 0.6500672101974487,
+    "eval_rewards/accuracies": 0.6031598448753357,
+    "eval_rewards/chosen": -1.0590553283691406,
+    "eval_rewards/margins": 0.1738797128200531,
+    "eval_rewards/rejected": -1.2329350709915161,
+    "eval_runtime": 384.1905,
+    "eval_samples": 4304,
+    "eval_samples_per_second": 11.203,
+    "eval_steps_per_second": 1.4
+}

runs/Jun09_13-29-09_poseidon/events.out.tfevents.1718026827.poseidon.4028099.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e3e2eb06afc3b1604098ec5ead94a82e2b77e4b32566de6a40be1ed815baa53
+size 841