Upload configs.yaml with huggingface_hub

configs.yaml

@@ -16,14 +16,14 @@ formatting: sharegpt
 global_batch_size: 96
 gradient_accumulation_steps: 3
 hub_model_id: neginr/multisubject_medicine_mc
-include_hp: dcft/train/hp_settings/paper/reasoning_small_OT7B.yaml
-learning_rate: 2.0e-07
+include_hp: dcft/train/hp_settings/paper/reasoning_small.yaml
+learning_rate: 2.0e-05
 logging_steps: 1
 lr_scheduler_type: cosine
 messages: conversations
-model_name_or_path: open-thoughts/OpenThinker2-7B
+model_name_or_path: Qwen/Qwen2.5-7B-Instruct
 num_train_epochs: 7.0
-output_dir: /scratch/08134/negin/dcft_checkpoints/r1_annotated_5k_OT7B_medicine
+output_dir: /scratch/08134/negin/dcft_checkpoints/r1_annotated_5k_medicine
 overwrite_cache: true
 per_device_train_batch_size: 1
 plot_loss: true
@@ -32,7 +32,7 @@ push_to_db: true
 push_to_hub: true
 report_to: wandb
 role_tag: from
-run_name: r1_annotated_5k_OT7B_medicine
+run_name: r1_annotated_5k_medicine
 save_strategy: epoch
 stage: sft
 template: qwen25