End of training

README.md

@@ -2,9 +2,13 @@
 license: apache-2.0
 base_model: ondevicellm/tinyllama_moe_sft_ultrachat200k_v2_epochs3
 tags:
+- alignment-handbook
+- generated_from_trainer
 - trl
 - dpo
 - generated_from_trainer
+datasets:
+- HuggingFaceH4/ultrafeedback_binarized
 model-index:
 - name: tinyllama_moe_dpo_ultrachat_v2_epochs3
   results: []
@@ -15,7 +19,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # tinyllama_moe_dpo_ultrachat_v2_epochs3
 
-This model is a fine-tuned version of [ondevicellm/tinyllama_moe_sft_ultrachat200k_v2_epochs3](https://huggingface.co/ondevicellm/tinyllama_moe_sft_ultrachat200k_v2_epochs3) on the None dataset.
+This model is a fine-tuned version of [ondevicellm/tinyllama_moe_sft_ultrachat200k_v2_epochs3](https://huggingface.co/ondevicellm/tinyllama_moe_sft_ultrachat200k_v2_epochs3) on the HuggingFaceH4/ultrafeedback_binarized dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.5855
 - Rewards/chosen: -0.9040

config.json

@@ -28,6 +28,6 @@
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
   "transformers_version": "4.36.2",
-  "use_cache": false,
+  "use_cache": true,
   "vocab_size": 32000
 }