Upload configs.yaml with huggingface_hub

configs.yaml

@@ -0,0 +1,43 @@
+adam_beta1: '0.9'
+adam_beta2: '0.999'
+bf16: 'True'
+cutoff_len: '2048'
+dataset: mlfoundations-dev/oh-dcft-v3.1-llama-3.1-405b
+dataset_dir: ONLINE
+ddp_timeout: '180000000'
+deepspeed: /opt/ml/code/zero3.json
+do_train: 'True'
+enable_liger_kernel: 'False'
+eval_strategy: epoch
+finetuning_type: full
+formatting: sharegpt
+global_batch_size: '512'
+gradient_accumulation_steps: '8'
+gradient_checkpointing: 'True'
+hub_model_id: mlfoundations-dev/mistral_7b_0-3_oh-dcft-v3.1-llama-3.1-405b-v2dummytesting
+learning_rate: 5e-06
+logging_steps: '10'
+lr_scheduler_kwargs: min_lr=5e-07
+lr_scheduler_type: cosine_with_min_lr
+max_grad_norm: '1'
+messages: conversations
+model_name_or_path: mistralai/Mistral-7B-v0.3
+neat_packing: 'True'
+num_train_epochs: '3.0'
+output_dir: /opt/ml/model
+overwrite_cache: 'True'
+overwrite_output_dir: 'True'
+packing: 'True'
+per_device_train_batch_size: '8'
+plot_loss: 'True'
+preprocessing_num_workers: '16'
+push_to_db: 'True'
+push_to_hub: 'True'
+report_to: wandb
+run_name: mistral_7b_0-3_oh-dcft-v3.1-llama-3.1-405b
+save_strategy: epoch
+stage: sft
+template: mistral
+val_size: '0.05'
+warmup_ratio: '0.05'
+weight_decay: '0.1'