End of training

README.md

@@ -7,14 +7,14 @@ tags:
 - generated_from_trainer
 base_model: MediaTek-Research/Breeze-7B-Instruct-v1_0
 model-index:
-- name: ROE_QA_Breeze-7B-Instruct-v1_0_Q30_80_20
+- name: ROE_QA_Breeze-7B-Instruct-v1_0_Q30_80_20_V2
   results: []
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 should probably proofread and complete it, then remove this comment. -->
 
-# ROE_QA_Breeze-7B-Instruct-v1_0_Q30_80_20
+# ROE_QA_Breeze-7B-Instruct-v1_0_Q30_80_20_V2
 
 This model is a fine-tuned version of [MediaTek-Research/Breeze-7B-Instruct-v1_0](https://huggingface.co/MediaTek-Research/Breeze-7B-Instruct-v1_0) on the None dataset.
 
@@ -35,15 +35,15 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 0.0002
+- learning_rate: 0.0001
 - train_batch_size: 2
 - eval_batch_size: 8
 - seed: 42
-- gradient_accumulation_steps: 4
-- total_train_batch_size: 8
+- gradient_accumulation_steps: 2
+- total_train_batch_size: 4
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
-- lr_scheduler_type: constant
-- lr_scheduler_warmup_ratio: 0.03
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_ratio: 0.05
 - num_epochs: 3
 - mixed_precision_training: Native AMP